Re: Problem with make
On Thu, Sep 22, 2011 at 11:52:36PM +0200, Reinhold Kainhofer wrote: > That's all just speculation, though, based on experience and gut feeling. Based on my spectulation and gut feeling from seeing this off and on for the past IIRC 1.5 years, it happens more often when you're building from an existing build dir (instead of a new one), and it happens more often the more -jX options you use. On my computer, with -j3 by default, I saw it something like 10% of the time when building from an existing build dir, and something like 2% of the time when building from a new build dir. Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
Am Thursday, 22. September 2011, 21:42:14 schrieb Peekay Ex: > if it is of any use, the last 3 patches I have just tested for David, > none gave me make failures at all. I even did two of them again twice, > just to see if it was a fluke. So it seems to be the 'type' of files > that get patched perhaps that generate this oddity? I would expect that the problem doesn't appear when no texinfo files are concerned. So pure C++ code changes shouldn't have any influence. On the other hand, it might be that either a new version number might cause such problems, or a rebuild due to changes in some texinfo file... That's all just speculation, though, based on experience and gut feeling. Cheers, Reinhold -- -- Reinhold Kainhofer, reinh...@kainhofer.com, http://reinhold.kainhofer.com/ * Financial & Actuarial Math., Vienna Univ. of Technology, Austria * http://www.fam.tuwien.ac.at/, DVR: 0005886 * LilyPond, Music typesetting, http://www.lilypond.org ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
Peekay Ex writes: > Hello, > > On Thu, Sep 22, 2011 at 6:49 PM, Graham Percival > wrote: >> On Thu, Sep 22, 2011 at 06:22:55PM +0100, Phil Holmes wrote: >>> - Original Message - From: "Graham Percival" >>> >>> >On Thu, Sep 22, 2011 at 10:17:50AM +0100, Phil Holmes wrote: >>> >>I'd like to add a note to this effect in the CG - anyone object? >>> > >>> >yes; I'd rather see an issue in the tracker instead. >>> >>> Can I suggest that this is rather overkill for an update to the CG? >>> I'd not disagree at all over notation ref, or code, but the CG isn't >>> supposed to be finely crafted words, it's supposed to be a quick >>> source of reference for contributors. You'll note that I thought >>> pushing directly to it was OK when I added stuff about regtest >>> comparisons. I think this is the same. >> >> oh, sorry, I was unclear. I don't want to see this bug documented >> in the CG; I want to see this bug fixed. It's going to be a 1-3 >> line fix to some build file. I'm optimistic that this can be >> achieved within a week. As such, I don't want an extra note in >> the CG that we'll have to remove in a few days. >> >> Cheers, >> - Graham >> > > if it is of any use, the last 3 patches I have just tested for David, > none gave me make failures at all. I even did two of them again twice, > just to see if it was a fluke. So it seems to be the 'type' of files > that get patched perhaps that generate this oddity? > > Reitveld issues 5023044, 5090045, 5083045. All of my patches tend to touch lily/parser.yy and lily/lexer.ll and not much else. -- David Kastrup ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
Hello, On Thu, Sep 22, 2011 at 6:49 PM, Graham Percival wrote: > On Thu, Sep 22, 2011 at 06:22:55PM +0100, Phil Holmes wrote: >> - Original Message - From: "Graham Percival" >> >> >On Thu, Sep 22, 2011 at 10:17:50AM +0100, Phil Holmes wrote: >> >>I'd like to add a note to this effect in the CG - anyone object? >> > >> >yes; I'd rather see an issue in the tracker instead. >> >> Can I suggest that this is rather overkill for an update to the CG? >> I'd not disagree at all over notation ref, or code, but the CG isn't >> supposed to be finely crafted words, it's supposed to be a quick >> source of reference for contributors. You'll note that I thought >> pushing directly to it was OK when I added stuff about regtest >> comparisons. I think this is the same. > > oh, sorry, I was unclear. I don't want to see this bug documented > in the CG; I want to see this bug fixed. It's going to be a 1-3 > line fix to some build file. I'm optimistic that this can be > achieved within a week. As such, I don't want an extra note in > the CG that we'll have to remove in a few days. > > Cheers, > - Graham > if it is of any use, the last 3 patches I have just tested for David, none gave me make failures at all. I even did two of them again twice, just to see if it was a fluke. So it seems to be the 'type' of files that get patched perhaps that generate this oddity? Reitveld issues 5023044, 5090045, 5083045. regards -- -- James ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
On Thu, Sep 22, 2011 at 06:22:55PM +0100, Phil Holmes wrote: > - Original Message - From: "Graham Percival" > > >On Thu, Sep 22, 2011 at 10:17:50AM +0100, Phil Holmes wrote: > >>I'd like to add a note to this effect in the CG - anyone object? > > > >yes; I'd rather see an issue in the tracker instead. > > Can I suggest that this is rather overkill for an update to the CG? > I'd not disagree at all over notation ref, or code, but the CG isn't > supposed to be finely crafted words, it's supposed to be a quick > source of reference for contributors. You'll note that I thought > pushing directly to it was OK when I added stuff about regtest > comparisons. I think this is the same. oh, sorry, I was unclear. I don't want to see this bug documented in the CG; I want to see this bug fixed. It's going to be a 1-3 line fix to some build file. I'm optimistic that this can be achieved within a week. As such, I don't want an extra note in the CG that we'll have to remove in a few days. Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
- Original Message - From: "Graham Percival" To: "Phil Holmes" Cc: ; "Devel" Sent: Thursday, September 22, 2011 5:35 PM Subject: Re: Problem with make On Thu, Sep 22, 2011 at 10:17:50AM +0100, Phil Holmes wrote: I'd like to add a note to this effect in the CG - anyone object? yes; I'd rather see an issue in the tracker instead. Can I suggest that this is rather overkill for an update to the CG? I'd not disagree at all over notation ref, or code, but the CG isn't supposed to be finely crafted words, it's supposed to be a quick source of reference for contributors. You'll note that I thought pushing directly to it was OK when I added stuff about regtest comparisons. I think this is the same. -- Phil Holmes ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
Graham Percival writes: > On Thu, Sep 22, 2011 at 11:44:29AM +0200, Reinhold Kainhofer wrote: >> >> Exactly. Make copies the *.texi file from the source to the build dir, runs >> extract-texi-filenames.py and then removes it again... > > Ouch. Is there any compelling reason to > 1) not run extract-texi-filename.py on the original file, > or > 2) not remove the file? I mean, it's not like a small text file > is going to change our 1.5 Gb build directory much. "not remove the file" does not help if the file is not properly in the dependency lists since then changes in one process are not synchronized with usage in another process. Temporary files can't properly be targets in a makefile. So they must not collide with the generation of other files and must be used exclusively in one rule, possibly by creating unique file names for them. -- David Kastrup ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
On Thu, Sep 22, 2011 at 11:44:29AM +0200, Reinhold Kainhofer wrote: > > Exactly. Make copies the *.texi file from the source to the build dir, runs > extract-texi-filenames.py and then removes it again... Ouch. Is there any compelling reason to 1) not run extract-texi-filename.py on the original file, or 2) not remove the file? I mean, it's not like a small text file is going to change our 1.5 Gb build directory much. Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
On Thu, Sep 22, 2011 at 10:17:50AM +0100, Phil Holmes wrote: > I'd like to add a note to this effect in the CG - anyone object? yes; I'd rather see an issue in the tracker instead. Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make on a commit *after* Davids last '\pushAtTag' commit - 264022bd6ebfed3220c0272d2c4a1c8ef9db4028
David Kastrup writes: > David Kastrup writes: > >> Peekay Ex writes: >> >>> Hello, >>> >>> I'm getting the following error when I tried to 'make' this morning. >>> >>> --snip-- >>> >>> /home/jlowe/lilypond-git/ly/string-tunings-init.ly:128:0: error: >>> syntax error, unexpected MUSIC_FUNCTION >> >> Am bisecting right now, this is likely caused by a change of mine I >> considered totally trivial. >> >> I know I know. > > Reverted. I have no clue whatsoever what could possibly cause this > error, probably something related to parser lookahead at inconvenient > times. > > It is always the trivial things that I mess up. Appalling oversight. composite_music could also have been prefix_composite_music, so replacing it with closed_music is not an option. I have to sort prefix_composite_music out into its components and try to come up with something that is at least as permissive as what we started with. And put up a patch review for the next "trivial" change I do to make somebody happy. -- David Kastrup ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make on a commit *after* Davids last '\pushAtTag' commit - 264022bd6ebfed3220c0272d2c4a1c8ef9db4028
David Kastrup writes: > Peekay Ex writes: > >> Hello, >> >> I'm getting the following error when I tried to 'make' this morning. >> >> --snip-- >> >> /home/jlowe/lilypond-git/ly/string-tunings-init.ly:128:0: error: >> syntax error, unexpected MUSIC_FUNCTION > > Am bisecting right now, this is likely caused by a change of mine I > considered totally trivial. > > I know I know. Reverted. I have no clue whatsoever what could possibly cause this error, probably something related to parser lookahead at inconvenient times. It is always the trivial things that I mess up. -- David Kastrup ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make on a commit *after* Davids last '\pushAtTag' commit - 264022bd6ebfed3220c0272d2c4a1c8ef9db4028
Peekay Ex writes: > Hello, > > I'm getting the following error when I tried to 'make' this morning. > > --snip-- > > /home/jlowe/lilypond-git/ly/string-tunings-init.ly:128:0: error: > syntax error, unexpected MUSIC_FUNCTION Am bisecting right now, this is likely caused by a change of mine I considered totally trivial. I know I know. -- David Kastrup ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make on a commit *after* Davids last '\pushAtTag' commit- 264022bd6ebfed3220c0272d2c4a1c8ef9db4028
On 22 September 2011 12:48, Phil Holmes wrote: > We probably need to learn how to use git bisect... It's David's most recent commit: 6c3445a0791831d450573cf583da36aecac5322c Cheers, Neil ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make on a commit *after* Davids last '\pushAtTag' commit- 264022bd6ebfed3220c0272d2c4a1c8ef9db4028
- Original Message - From: "Peekay Ex" To: "Devel" Sent: Thursday, September 22, 2011 12:36 PM Subject: Problem with make on a commit *after* Davids last '\pushAtTag' commit- 264022bd6ebfed3220c0272d2c4a1c8ef9db4028 Hello, I'm getting the following error when I tried to 'make' this morning. --snip-- [snip] so somewhere after that is where the problem occured. I've gone back through the commits but cannot see anything specific that changes 'string-tunings-init.ly' but I see nothing. It was from a newly created out-of-tree build. Can someone check this? Checked and confirmed. Fresh build directory, new pull. We probably need to learn how to use git bisect... -- Phil Holmes ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Problem with make on a commit *after* Davids last '\pushAtTag' commit - 264022bd6ebfed3220c0272d2c4a1c8ef9db4028
Hello, I'm getting the following error when I tried to 'make' this morning. --snip-- /home/jlowe/lilypond-git/ly/string-tunings-init.ly:128:0: error: syntax error, unexpected MUSIC_FUNCTION \makeDefaultStringTunings #defaultStringTunings ] [/home/jlowe/lilypond-git/ly/property-init.ly] [/home/jlowe/lilypond-git/ly/grace-init.ly] [/home/jlowe/lilypond-git/ly/midi-init.ly [/home/jlowe/lilypond-git/ly/performer-init.ly]] [/home/jlowe/lilypond-git/ly/paper-defaults-init.ly [/home/jlowe/lilypond-git/ly/titling-init.ly]] [/home/jlowe/lilypond-git/ly/context-mods-init.ly] [/home/jlowe/lilypond-git/ly/engraver-init.lyBacktrace: In unknown file: ?: 0* [lilypond-main ("/home/jlowe/lilypond-git/ly/generate-documentation")] In /home/jlowe/lilypond-git/build/out/share/lilypond/current/scm/lily.scm: 816: 1* (let* ((failed #)) (if (ly:get-option #) (begin #)) ...) 816: 2* [lilypond-all ("/home/jlowe/lilypond-git/ly/generate-documentation")] 829: 3 (let* ((failed #) (separate-logs #) (ping-log #) ...) (gc) ...) 840: 4* [for-each # #] In unknown file: ?: 5* [# "/home/jlowe/lilypond-git/ly/generate-documentation"] In /home/jlowe/lilypond-git/build/out/share/lilypond/current/scm/lily.scm: 842: 6* (let* (# # #) (if separate-logs #) (if ping-log #) ...) 853: 7* [lilypond-file # ...] 879: 8 [catch ly-file-failed # #] In unknown file: ?: 9* [#] In /home/jlowe/lilypond-git/build/out/share/lilypond/current/scm/lily.scm: 880: 10* [ly:parse-file "/home/jlowe/lilypond-git/ly/generate-documentation"] /home/jlowe/lilypond-git/build/out/share/lilypond/current/scm/lily.scm:880:21: In procedure ly:parse-file in expression (ly:parse-file file-name): /home/jlowe/lilypond-git/build/out/share/lilypond/current/scm/lily.scm:880:21: Unbound variable: guitar-tuning make[1]: *** [out/internals.texi] Error 1 make[1]: Leaving directory `/home/jlowe/lilypond-git/build/Documentation' make: *** [all] Error 2 jlowe@jlowe-lilydevVM:~/lilypond-git/build$ --snip-- The last successful make I did was for Davids last '\pushAtTag' commit - 264022bd6ebfed3220c0272d2c4a1c8ef9db4028 so somewhere after that is where the problem occured. I've gone back through the commits but cannot see anything specific that changes 'string-tunings-init.ly' but I see nothing. It was from a newly created out-of-tree build. Can someone check this? -- -- James ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
Hello, On Thu, Sep 22, 2011 at 10:21 AM, Phil Holmes wrote: > - Original Message - From: "Reinhold Kainhofer" > > To: > Cc: "David Kastrup" > Sent: Thursday, September 22, 2011 2:07 AM > Subject: Re: Problem with make > > >> Am Thursday, 22. September 2011, 01:55:15 schrieb David Kastrup: >>> >>> > yep, happens about 10% of the time for me. Running "make" again >>> > fixes it. (almost always -- the chance of two failing runs is >>> > about 1%. That's happened twice to me that I can recall) >>> > - - - >>> > >>> > happens to me nearly every time i make. You get news.tely and >>> > authors.texi. Failing. Run make again and it's fine. Never had two on >>> > the trot like graham. >> >> [..] >>> >>> That points to either a problem with parallel make processes, or more >>> likely a time stamp resolution problem. >> >> I only get such a problem if I call "make -j3" (or something other >1). I >> have >> never experienced it with "make", so my guess would be that the >> dependencies >> of our texinfo files are not complete. >> >> Cheers, >> Reinhold > > > I was getting it when I went from make -j9 to make -j5, but didn't want to > be bothered with running a single core. I'm assuming you're correct, > though. > No. It does happen with just a make (I use a single core VM at work for Simple testing) just not very often. I also find that when I am testing patches that there are some patches that never give me this problem, but I've never been able to figure it out (for example if it is just *.cc files patches or *.ly files or some other combination) to give me any more clues. I can start to look a bit harder if people want or am happy to test theories as I, like phil, have a powerful machine so it is only a few minutes. -- -- James ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
Am Thursday, 22. September 2011, 09:58:05 schrieb David Kastrup: > Reinhold Kainhofer writes: > > I only get such a problem if I call "make -j3" (or something other > > >1). I have never experienced it with "make", so my guess would be > > that the dependencies of our texinfo files are not complete. > > That would probably mean that one make process determines that a > dependency is available and up to date and proceeds on this assumption, > and then another make process removes that file as a sideeffect of > building a different target before the first make process has finished > working with the file. Exactly. Make copies the *.texi file from the source to the build dir, runs extract-texi-filenames.py and then removes it again... I haven't looked into which stepmake snippet has these rules, but that's what I understand from the output of make. Cheers, Reinhold -- -- Reinhold Kainhofer, reinh...@kainhofer.com, http://reinhold.kainhofer.com/ * Financial & Actuarial Math., Vienna Univ. of Technology, Austria * http://www.fam.tuwien.ac.at/, DVR: 0005886 * LilyPond, Music typesetting, http://www.lilypond.org ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
- Original Message - From: "Reinhold Kainhofer" To: Cc: "David Kastrup" Sent: Thursday, September 22, 2011 2:07 AM Subject: Re: Problem with make Am Thursday, 22. September 2011, 01:55:15 schrieb David Kastrup: > yep, happens about 10% of the time for me. Running "make" again > fixes it. (almost always -- the chance of two failing runs is > about 1%. That's happened twice to me that I can recall) > - - - > > happens to me nearly every time i make. You get news.tely and > authors.texi. Failing. Run make again and it's fine. Never had two on > the trot like graham. [..] That points to either a problem with parallel make processes, or more likely a time stamp resolution problem. I only get such a problem if I call "make -j3" (or something other >1). I have never experienced it with "make", so my guess would be that the dependencies of our texinfo files are not complete. Cheers, Reinhold I was getting it when I went from make -j9 to make -j5, but didn't want to be bothered with running a single core. I'm assuming you're correct, though. -- Phil Holmes ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
- Original Message - From: "David Kastrup" To: Sent: Thursday, September 22, 2011 12:59 AM Subject: Re: Problem with make David Kastrup writes: "pkx1...@gmail.com" writes: hello, On Wed, Sep 21, 2011 at 05:13:00PM +0100, Phil Holmes wrote: On my fast build system, I can't currently get a successful make. Abort changes, pull, clean build directory. The build ends with: ... As you see, the problem is a missing AUTHORS.texi. The odd thing is that on previous make runs, I get yep, happens about 10% of the time for me. Running "make" again fixes it. (almost always -- the chance of two failing runs is about 1%. That's happened twice to me that I can recall) - - - happens to me nearly every time i make. You get news.tely and authors.texi. Failing. Run make again and it's fine. Never had two on the trot like graham. I probably run make about 10 times a day at the moment, checking patches. I'm used to it and just assumed it was one of our build quirks. You'll see lots more on faster machines in my own personal experience. That points to either a problem with parallel make processes, or more likely a time stamp resolution problem. When file modification dates are only accessed with second resolution (because the info is not available in the file system type, or the application does not use it) and a process for updates is quite fast, an updated dependent file may seem to have the same time stamp as its original. Expounding on that theory and doing pattern matching: does the problem get better or worse when you replace the > in python/book_snippets.py:781: > os.stat (single)[stat.ST_MTIME]))): with >= ? I'll have a look. It may have nothing whatsoever to do with your problem, but that's a reference to a modification time I can see in the code. And what file system do you have? fat32 does not support more than second resolution IIRC. Whatever Ubuntu uses as a default. On my windows systems I always use NTFS. -- Phil Holmes ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
I'd like to add a note to this effect in the CG - anyone object? -- Phil Holmes - Original Message - From: To: "Phil Holmes" ; "Graham Percival" Cc: "Devel" Sent: Thursday, September 22, 2011 12:05 AM Subject: RE: Problem with make hello, On Wed, Sep 21, 2011 at 05:13:00PM +0100, Phil Holmes wrote: On my fast build system, I can't currently get a successful make. Abort changes, pull, clean build directory. The build ends with: ... As you see, the problem is a missing AUTHORS.texi. The odd thing is that on previous make runs, I get yep, happens about 10% of the time for me. Running "make" again fixes it. (almost always -- the chance of two failing runs is about 1%. That's happened twice to me that I can recall) - - - happens to me nearly every time i make. You get news.tely and authors.texi. Failing. Run make again and it's fine. Never had two on the trot like graham. I probably run make about 10 times a day at the moment, checking patches. I'm used to it and just assumed it was one of our build quirks. You'll see lots more on faster machines in my own personal experience. James ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
Reinhold Kainhofer writes: > Am Thursday, 22. September 2011, 01:55:15 schrieb David Kastrup: >> > yep, happens about 10% of the time for me. Running "make" again >> > fixes it. (almost always -- the chance of two failing runs is >> > about 1%. That's happened twice to me that I can recall) >> > - - - >> > >> > happens to me nearly every time i make. You get news.tely and >> > authors.texi. Failing. Run make again and it's fine. Never had two on >> > the trot like graham. > [..] >> That points to either a problem with parallel make processes, or more >> likely a time stamp resolution problem. > > I only get such a problem if I call "make -j3" (or something other > >1). I have never experienced it with "make", so my guess would be > that the dependencies of our texinfo files are not complete. That would probably mean that one make process determines that a dependency is available and up to date and proceeds on this assumption, and then another make process removes that file as a sideeffect of building a different target before the first make process has finished working with the file. -- David Kastrup ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
On 11-09-21 11:06 AM, Graham Percival wrote: On Wed, Sep 21, 2011 at 05:13:00PM +0100, Phil Holmes wrote: On my fast build system, I can't currently get a successful make. Abort changes, pull, clean build directory. The build ends with: ... As you see, the problem is a missing AUTHORS.texi. The odd thing is that on previous make runs, I get yep, happens about 10% of the time for me. Running "make" again fixes it. (almost always -- the chance of two failing runs is about 1%. That's happened twice to me that I can recall) Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel I had the problem a day or so ago, Phil. I ended up completely nuking /lilypond-git and using lily-git to restart. I didn't have a recent enough tarball to be worth simply copying the file over, as a test. The nuke to bare metal seems to have done it. The truly wierd thing, now that I recall a bit more in detail, was that I couldn't find a trace of AUTHORS.texi using git cola or gitg. Of course, I may have been searching for a misspelling, but even browsing in tree mode didn't show me anything. Cheers, Colin -- I've learned that you shouldn't go through life with a catcher's mitt on both hands. You need to be able to throw something back. -Maya Angelou, poet (1928- ) ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
On Thu, Sep 22, 2011 at 01:59:53AM +0200, David Kastrup wrote: > David Kastrup writes: > > > "pkx1...@gmail.com" writes: > > > >> hello, > >> > >> On Wed, Sep 21, 2011 at 05:13:00PM +0100, Phil Holmes wrote: > >>> On my fast build system, I can't currently get a successful make. > >>> Abort changes, pull, clean build directory. The build ends with: > >> ... > >>> As you see, the problem is a missing AUTHORS.texi. The odd thing is > >>> that on previous make runs, I get > > Expounding on that theory and doing pattern matching: does the problem > get better or worse when you replace the > in > python/book_snippets.py:781: > os.stat > (single)[stat.ST_MTIME]))): > > with >= ? I doubt that's the issue; AUTHORS.texi has nothing to do with lilypond-book. We do something weird and complicated to build the TOPDOC_FILES. I've tried poking around every so often, but I always get lost around the third or fourth redirect. If anybody wants to go spelunking, start off with git grep AUTHORS and pay particular attention to any GNUmakefile or anything in make/ or stepmake/ or stepmake/stepmake/ . Sooner or later, you'll find out exactly how AUTHORS.texi is built, but I would budget at least an hour for grepping and reading various build files. Also, keeping notes with pen and paper might not be a bad idea. Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
Am Thursday, 22. September 2011, 01:55:15 schrieb David Kastrup: > > yep, happens about 10% of the time for me. Running "make" again > > fixes it. (almost always -- the chance of two failing runs is > > about 1%. That's happened twice to me that I can recall) > > - - - > > > > happens to me nearly every time i make. You get news.tely and > > authors.texi. Failing. Run make again and it's fine. Never had two on > > the trot like graham. [..] > That points to either a problem with parallel make processes, or more > likely a time stamp resolution problem. I only get such a problem if I call "make -j3" (or something other >1). I have never experienced it with "make", so my guess would be that the dependencies of our texinfo files are not complete. Cheers, Reinhold -- -- Reinhold Kainhofer, reinh...@kainhofer.com, http://reinhold.kainhofer.com/ * Financial & Actuarial Math., Vienna Univ. of Technology, Austria * http://www.fam.tuwien.ac.at/, DVR: 0005886 * LilyPond, Music typesetting, http://www.lilypond.org ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
David Kastrup writes: > "pkx1...@gmail.com" writes: > >> hello, >> >> On Wed, Sep 21, 2011 at 05:13:00PM +0100, Phil Holmes wrote: >>> On my fast build system, I can't currently get a successful make. >>> Abort changes, pull, clean build directory. The build ends with: >> ... >>> As you see, the problem is a missing AUTHORS.texi. The odd thing is >>> that on previous make runs, I get >> >> yep, happens about 10% of the time for me. Running "make" again >> fixes it. (almost always -- the chance of two failing runs is >> about 1%. That's happened twice to me that I can recall) >> - - - >> >> happens to me nearly every time i make. You get news.tely and >> authors.texi. Failing. Run make again and it's fine. Never had two on >> the trot like graham. >> >> I probably run make about 10 times a day at the moment, checking >> patches. I'm used to it and just assumed it was one of our build >> quirks. You'll see lots more on faster machines in my own personal >> experience. > > That points to either a problem with parallel make processes, or more > likely a time stamp resolution problem. When file modification dates > are only accessed with second resolution (because the info is not > available in the file system type, or the application does not use it) > and a process for updates is quite fast, an updated dependent file may > seem to have the same time stamp as its original. Expounding on that theory and doing pattern matching: does the problem get better or worse when you replace the > in python/book_snippets.py:781: > os.stat (single)[stat.ST_MTIME]))): with >= ? It may have nothing whatsoever to do with your problem, but that's a reference to a modification time I can see in the code. And what file system do you have? fat32 does not support more than second resolution IIRC. -- David Kastrup ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
"pkx1...@gmail.com" writes: > hello, > > On Wed, Sep 21, 2011 at 05:13:00PM +0100, Phil Holmes wrote: >> On my fast build system, I can't currently get a successful make. >> Abort changes, pull, clean build directory. The build ends with: > ... >> As you see, the problem is a missing AUTHORS.texi. The odd thing is >> that on previous make runs, I get > > yep, happens about 10% of the time for me. Running "make" again > fixes it. (almost always -- the chance of two failing runs is > about 1%. That's happened twice to me that I can recall) > - - - > > happens to me nearly every time i make. You get news.tely and > authors.texi. Failing. Run make again and it's fine. Never had two on > the trot like graham. > > I probably run make about 10 times a day at the moment, checking > patches. I'm used to it and just assumed it was one of our build > quirks. You'll see lots more on faster machines in my own personal > experience. That points to either a problem with parallel make processes, or more likely a time stamp resolution problem. When file modification dates are only accessed with second resolution (because the info is not available in the file system type, or the application does not use it) and a process for updates is quite fast, an updated dependent file may seem to have the same time stamp as its original. -- David Kastrup ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
RE: Problem with make
hello, On Wed, Sep 21, 2011 at 05:13:00PM +0100, Phil Holmes wrote: > On my fast build system, I can't currently get a successful make. > Abort changes, pull, clean build directory. The build ends with: ... > As you see, the problem is a missing AUTHORS.texi. The odd thing is > that on previous make runs, I get yep, happens about 10% of the time for me. Running "make" again fixes it. (almost always -- the chance of two failing runs is about 1%. That's happened twice to me that I can recall) - - - happens to me nearly every time i make. You get news.tely and authors.texi. Failing. Run make again and it's fine. Never had two on the trot like graham. I probably run make about 10 times a day at the moment, checking patches. I'm used to it and just assumed it was one of our build quirks. You'll see lots more on faster machines in my own personal experience. James ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: Problem with make
On Wed, Sep 21, 2011 at 05:13:00PM +0100, Phil Holmes wrote: > On my fast build system, I can't currently get a successful make. > Abort changes, pull, clean build directory. The build ends with: ... > As you see, the problem is a missing AUTHORS.texi. The odd thing is > that on previous make runs, I get yep, happens about 10% of the time for me. Running "make" again fixes it. (almost always -- the chance of two failing runs is about 1%. That's happened twice to me that I can recall) Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel
Problem with make
On my fast build system, I can't currently get a successful make. Abort changes, pull, clean build directory. The build ends with: make[2]: Entering directory `/media/IntelSSD/lilypond/lilypond-git/build/Documentation/topdocs' LILYPOND_VERSION=2.15.13 [snip options] out/NEWS.tely /usr/bin/python /home/phil/lilypond-git/scripts/build/create-weblinks-itexi.py > out/weblinks.itexi langdefs.py: warning: lilypond-doc gettext domain not found. langdefs.py: warning: lilypond-doc gettext domain not found. LANG= makeinfo --enable-encoding -I /home/phil/lilypond-git/Documentation -I /home/phil/lilypond-git/Documentation/usage -I /home/phil/lilypond-git/Documentation/contributor -I/home/phil/lilypond-git/Documentation/topdocs -I./out --no-split --no-headers --output out/AUTHORS.txt out/AUTHORS.texi out/AUTHORS.texi: No such file or directory Reading out/NEWS.tely... make[2]: *** [out/AUTHORS.txt] Error 1 make[2]: *** Waiting for unfinished jobs Running texi2pdf on file /tmp/tmpCitEBl.texi to detect default page settings. Executing: LC_ALL=C texi2pdf -c -o /tmp/tmpCitEBl.pdf /tmp/tmpCitEBl.texi Auto-detected values are: {'exampleindent': '10.16\\mm', 'line-width': '160\\mm'} Dissecting... Writing snippets... All snippets are up to date... Compiling /media/IntelSSD/lilypond/lilypond-git/build/Documentation/topdocs/out/NEWS.texi... Writing `/media/IntelSSD/lilypond/lilypond-git/build/Documentation/topdocs/out/NEWS.texi'... Processing include: macros.itexi Reading /home/phil/lilypond-git/Documentation/macros.itexi... Dissecting... Writing snippets... All snippets are up to date... Compiling /media/IntelSSD/lilypond/lilypond-git/build/Documentation/topdocs/out/macros.texi... /media/IntelSSD/lilypond/lilypond-git/build/Documentation/topdocs/out/macros.texi is up to date. Processing include: version.itexi Reading /media/IntelSSD/lilypond/lilypond-git/build/Documentation/topdocs/out/version.itexi... Dissecting... Writing snippets... All snippets are up to date... Compiling /media/IntelSSD/lilypond/lilypond-git/build/Documentation/topdocs/out/version.texi... /media/IntelSSD/lilypond/lilypond-git/build/Documentation/topdocs/out/version.texi is up to date. Processing include: common-macros.itexi Reading /home/phil/lilypond-git/Documentation/common-macros.itexi... Dissecting... Writing snippets... All snippets are up to date... Compiling /media/IntelSSD/lilypond/lilypond-git/build/Documentation/topdocs/out/common-macros.texi... /media/IntelSSD/lilypond/lilypond-git/build/Documentation/topdocs/out/common-macros.texi is up to date. lilypond-book.py (GNU LilyPond) 2.15.13 make[2]: *** wait: No child processes. Stop. make[1]: *** [all] Error 2 rm out/weblinks.itexi make[1]: Leaving directory `/media/IntelSSD/lilypond/lilypond-git/build/Documentation' make: *** [all] Error 2 As you see, the problem is a missing AUTHORS.texi. The odd thing is that on previous make runs, I get make[2]: Entering directory `/media/IntelSSD/lilypond/lilypond-git/build/Documentation/topdocs' (as above), but nothing is built - so I'm not sure why it's suddenly decided to build NEWS.tely. Is anyone else seeing this? Anyone aware of why this is happening? -- Phil Holmes ___ lilypond-devel mailing list lilypond-devel@gnu.org https://lists.gnu.org/mailman/listinfo/lilypond-devel