apenwarr-redo/redo-log.py

241 lines
7.7 KiB
Python
Raw Normal View History

redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
#!/usr/bin/env python
import errno, fcntl, os, re, struct, sys, termios, time
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
import options
optspec = """
redo-log [options...] [targets...]
--
r,recursive show build logs for dependencies too
u,unchanged show lines for dependencies not needing to be rebuilt
f,follow keep watching for more lines to be appended (like tail -f)
no-details only show 'redo' recursion trace, not build output
no-colorize don't colorize 'redo' log messages
no-status don't display build summary line in --follow
raw-logs don't format logs, just send raw output straight to stdout
debug-locks print messages about file locking (useful for debugging)
debug-pids print process ids in log messages (useful for debugging)
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
ack-fd= (internal use only) print REDO-OK to this fd upon starting
"""
o = options.Options(optspec)
(opt, flags, extra) = o.parse(sys.argv[1:])
targets = extra
import vars_init
vars_init.init(list(targets))
import vars, logs, state
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
already = set()
queue = []
depth = []
total_lines = 0
status = None
# regexp for matching "redo" lines in the log, which we use for recursion.
# format:
# redo path/to/target which might have spaces
# redo [unchanged] path/to/target which might have spaces
# redo path/to/target which might have spaces (comment)
REDO_LINE_RE = re.compile(r'^@@REDO:([^@]+)@@ (.*)\n$')
def _atoi(s):
try:
return int(s)
except TypeError:
return 0
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
def _tty_width():
s = struct.pack("HHHH", 0, 0, 0, 0)
try:
import fcntl, termios
s = fcntl.ioctl(sys.stderr.fileno(), termios.TIOCGWINSZ, s)
except (IOError, ImportError):
return _atoi(os.environ.get('WIDTH')) or 70
(ysize,xsize,ypix,xpix) = struct.unpack('HHHH', s)
return xsize or 70
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
def is_locked(fid):
return (fid is not None) and not state.Lock(fid=fid).trylock()
def _fix_depth():
vars.DEPTH = len(depth) * ' '
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
def catlog(t):
global total_lines, status
if t in already:
return
if t != '-':
depth.append(t)
_fix_depth()
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
already.add(t)
if t == '-':
f = sys.stdin
fid = None
loglock = None
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
logname = None
else:
try:
sf = state.File(name=t, allow_add=False)
except KeyError:
sys.stderr.write('redo-log: %r: not known to redo.\n' % (t,))
sys.exit(24)
fid = sf.id
del sf
state.rollback()
logname = state.logname(fid)
loglock = state.Lock(fid + state.LOG_LOCK_MAGIC)
loglock.waitlock(shared=True)
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
f = None
delay = 0.01
was_locked = is_locked(fid)
line_head = ''
width = _tty_width()
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
while 1:
if not f:
try:
f = open(logname)
except IOError, e:
if e.errno == errno.ENOENT:
# ignore files without logs
pass
else:
raise
if f:
# Note: normally includes trailing \n.
# In 'follow' mode, might get a line with no trailing \n
# (eg. when ./configure is halfway through a test), which we
# deal with below.
line = f.readline()
else:
line = None
if not line and (not opt.follow or not was_locked):
# file not locked, and no new lines: done
break
if not line:
was_locked = is_locked(fid)
if opt.follow:
if opt.status:
width = _tty_width()
head = 'redo %s ' % ('{:,}'.format(total_lines))
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
tail = ''
for n in reversed(depth):
remain = width - len(head) - len(tail)
# always leave room for a final '... ' prefix
if remain < len(n) + 4 + 1 or remain <= 4:
if len(n) < 6 or remain < 6 + 1 + 4:
tail = '... %s' % tail
else:
start = len(n) - (remain - 3 - 1)
tail = '...%s %s' % (n[start:], tail)
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
break
elif n != '-':
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
tail = n + ' ' + tail
status = head + tail
if len(status) > width:
sys.stderr.write('\nOVERSIZE STATUS (%d):\n%r\n' %
(len(status), status))
assert(len(status) <= width)
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
sys.stdout.flush()
sys.stderr.write('\r%-*.*s\r' % (width, width, status))
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
time.sleep(min(delay, 1.0))
delay += 0.01
continue
total_lines += 1
delay = 0.01
if not line.endswith('\n'):
line_head += line
continue
if line_head:
line = line_head + line
line_head = ''
if status:
sys.stdout.flush()
sys.stderr.write('\r%-*.*s\r' % (width, width, ''))
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
status = None
g = re.match(REDO_LINE_RE, line)
if g:
# FIXME: print prefix if @@REDO is not at start of line.
# logs.PrettyLog does it, but only if we actually call .write().
words, text = g.groups()
kind, pid, when = words.split(':')[0:3]
if kind == 'unchanged':
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
if opt.unchanged:
if opt.debug_locks:
logs.write(line.rstrip())
elif text not in already:
logs.meta('do', text)
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
if opt.recursive:
if loglock: loglock.unlock()
catlog(text)
if loglock: loglock.waitlock(shared=True)
elif kind in ('do', 'waiting', 'locked', 'unlocked'):
if opt.debug_locks:
logs.write(line.rstrip())
elif text not in already:
logs.meta('do', text)
if opt.recursive:
assert text
if loglock: loglock.unlock()
catlog(text)
if loglock: loglock.waitlock(shared=True)
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
else:
logs.write(line.rstrip())
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
else:
if opt.details:
logs.write(line.rstrip())
if loglock:
loglock.unlock()
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
if status:
sys.stdout.flush()
sys.stderr.write('\r%-*.*s\r' % (width, width, ''))
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
status = None
if line_head:
# partial line never got terminated
print line_head
if t != '-':
assert(depth[-1] == t)
depth.pop(-1)
_fix_depth()
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
try:
if not targets:
sys.stderr.write('redo-log: give at least one target; maybe "all"?\n')
sys.exit(1)
if opt.status < 2 and not os.isatty(2):
opt.status = False
if opt.raw_logs:
logs.setup(file=sys.stdout, pretty=False)
else:
logs.setup(file=sys.stdout, pretty=True)
if opt.debug_locks:
vars.DEBUG_LOCKS = 1
if opt.debug_pids:
vars.DEBUG_PIDS = 1
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
if opt.ack_fd:
# Write back to owner, to let them know we started up okay and
# will be able to see their error output, so it's okay to close
# their old stderr.
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
ack_fd = int(opt.ack_fd)
assert(ack_fd > 2)
if os.write(ack_fd, 'REDO-OK\n') != 8:
raise Exception('write to ack_fd returned wrong length')
os.close(ack_fd)
queue += targets
while queue:
t = queue.pop(0)
if t != '-':
logs.meta('do', t)
redo-log: capture and linearize the output of redo builds. redo now saves the stderr from every .do script, for every target, into a file in the .redo directory. That means you can look up the logs from the most recent build of any target using the new redo-log command, for example: redo-log -r all The default is to show logs non-recursively, that is, it'll show when a target does redo-ifchange on another target, but it won't recurse into the logs for the latter target. With -r (recursive), it does. With -u (unchanged), it does even if redo-ifchange discovered that the target was already up-to-date; in that case, it prints the logs of the *most recent* time the target was generated. With --no-details, redo-log will show only the 'redo' lines, not the other log messages. For very noisy build systems (like recursing into a 'make' instance) this can be helpful to get an overview of what happened, without all the cruft. You can use the -f (follow) option like tail -f, to follow a build that's currently in progress until it finishes. redo itself spins up a copy of redo-log -r -f while it runs, so you can see what's going on. Still broken in this version: - No man page or new tests yet. - ANSI colors don't yet work (unless you use --raw-logs, which gives the old-style behaviour). - You can't redirect the output of a sub-redo to a file or a pipe right now, because redo-log is eating it. - The regex for matching 'redo' lines in the log is very gross. Instead, we should put the raw log files in a more machine-parseable format, and redo-log should turn that into human-readable format. - redo-log tries to "linearize" the logs, which makes them comprehensible even for a large parallel build. It recursively shows log messages for each target in depth-first tree order (by tracing into a new target every time it sees a 'redo' line). This works really well, but in some specific cases, the "topmost" redo instance can get stuck waiting for a jwack token, which makes it look like the whole build has stalled, when really redo-log is just waiting a long time for a particular subprocess to be able to continue. We'll need to add a specific workaround for that.
2018-11-03 22:09:18 -04:00
catlog(t)
except KeyboardInterrupt:
sys.exit(200)
except IOError, e:
if e.errno == errno.EPIPE:
pass
else:
raise