Re: cvs, stop ignoring *@#%@# files/directories named core
On 2012/03/23 11:33, Marc Espie wrote: On Tue, Mar 20, 2012 at 11:42:23AM +0100, Mark Kettenis wrote: Date: Tue, 20 Mar 2012 10:07:35 + From: Stuart Henderson s...@spacehopper.org core dumps on OpenBSD are not named 'core' anyway so ignoring them is pointless and gets in the way. OK? Why should cvs on OpenBSD behave different from cvs on other platforms? Because it already does. We have already adapted it to current practices, such as ignoring vim .swp files. Do you want this to be reversed as well ? People should not name files or directories core. It leads to accidents like sysadmins accidentally removing stuff when trying to free up disk space by removing all core files. Nobody is free from idiots. Fortunately, most idiots don't run OpenBSD. And even then, idiots won't remove core directories, since rm no longer works on directories... Seriously, I really think sthen@'s change should go in. 1/ It reflects how our system works. 2/ Who uses cvs on other platforms anyways ? 3/ This is no place for a holier than thou dissertation. If people want to name files OR DIRECTORIES core, so what ? It's THEIR problem. 4/ The current setup of cvs already causes problems, as we REGULARLY have software that's organized as core plugins. I see absolutely NO ISSUE in having directories named core. An addition to ignore .git files, some of us use git locally to track changes pre-commit and, just like the existing ignore entries for RCS/SCCS/CVS control files, there is no reason to import these to CVS without an explicit ignore. Index: src/ignore.c === RCS file: /cvs/src/gnu/usr.bin/cvs/src/ignore.c,v retrieving revision 1.16 diff -u -p -r1.16 ignore.c --- src/ignore.c4 Mar 2012 04:05:15 - 1.16 +++ src/ignore.c26 Mar 2012 09:11:41 - @@ -33,8 +33,8 @@ static int ign_size; /* This many slot static int ign_hold = -1; /* Index where first temporary item * is held */ -const char *ign_default = . .. core RCSLOG tags TAGS RCS SCCS .make.state\ - .*.swp\ +const char *ign_default = . .. RCSLOG tags TAGS RCS SCCS .make.state\ + .*.swp *.core .git\ .nse_depinfo #* .#* cvslog.* ,* CVS CVS.adm .del-* *.a *.olb *.o *.obj\ *.so *.Z *~ *.old *.elc *.ln *.bak *.BAK *.orig *.rej *.exe _$* *$ *.depend;
Re: cvs, stop ignoring *@#%@# files/directories named core
Stuart Henderson wrote [2012-03-26 11:16+0200]: An addition to ignore .git files, some of us use git locally to track changes pre-commit and, just like the existing ignore entries for RCS/SCCS/CVS control files, there is no reason to import these to CVS without an explicit ignore. Yeah! Adding .gitignore in addition would be fine, then. --steffen Forza Figa!
Re: cvs, stop ignoring *@#%@# files/directories named core
On 2012/03/26 10:16, Stuart Henderson wrote: An addition to ignore .git files, some of us use git locally to track changes pre-commit and, just like the existing ignore entries for RCS/SCCS/CVS control files, there is no reason to import these to CVS without an explicit ignore. A few offlist comments asked about adding ignores for other vcs. I think this makes sense for vcs which people actually use over the top of cvs but I don't know what people are actually using for this, and I'm not sure it makes sense to add a bunch pre-emptively. (bzr, hg, git, svn, fossil, where do we stop?) So what do people use with these directories? (This isn't a general what vcs do people use question, it refers to what people use in the *same directories* that they might also use with cvs, and this is primarily a problem with cvs import mainly used by people committing new ports and merging upstream code into the src and xenocara trees). Index: src/ignore.c === RCS file: /cvs/src/gnu/usr.bin/cvs/src/ignore.c,v retrieving revision 1.16 diff -u -p -r1.16 ignore.c --- src/ignore.c 4 Mar 2012 04:05:15 - 1.16 +++ src/ignore.c 26 Mar 2012 09:11:41 - @@ -33,8 +33,8 @@ static int ign_size;/* This many slot static int ign_hold = -1;/* Index where first temporary item * is held */ -const char *ign_default = . .. core RCSLOG tags TAGS RCS SCCS .make.state\ - .*.swp\ +const char *ign_default = . .. RCSLOG tags TAGS RCS SCCS .make.state\ + .*.swp *.core .git\ .nse_depinfo #* .#* cvslog.* ,* CVS CVS.adm .del-* *.a *.olb *.o *.obj\ *.so *.Z *~ *.old *.elc *.ln *.bak *.BAK *.orig *.rej *.exe _$* *$ *.depend;
Re: cvs, stop ignoring *@#%@# files/directories named core
On Tue, Mar 20, 2012 at 11:42:23AM +0100, Mark Kettenis wrote: Date: Tue, 20 Mar 2012 10:07:35 + From: Stuart Henderson s...@spacehopper.org core dumps on OpenBSD are not named 'core' anyway so ignoring them is pointless and gets in the way. OK? Why should cvs on OpenBSD behave different from cvs on other platforms? Because it already does. We have already adapted it to current practices, such as ignoring vim .swp files. Do you want this to be reversed as well ? People should not name files or directories core. It leads to accidents like sysadmins accidentally removing stuff when trying to free up disk space by removing all core files. Nobody is free from idiots. Fortunately, most idiots don't run OpenBSD. And even then, idiots won't remove core directories, since rm no longer works on directories... Seriously, I really think sthen@'s change should go in. 1/ It reflects how our system works. 2/ Who uses cvs on other platforms anyways ? 3/ This is no place for a holier than thou dissertation. If people want to name files OR DIRECTORIES core, so what ? It's THEIR problem. 4/ The current setup of cvs already causes problems, as we REGULARLY have software that's organized as core plugins. I see absolutely NO ISSUE in having directories named core.
Re: cvs, stop ignoring *@#%@# files/directories named core
On Tue, Mar 20, 2012 at 10:07:35AM +, Stuart Henderson wrote: core dumps on OpenBSD are not named 'core' anyway so ignoring them is pointless and gets in the way. Why not making it '*.core' so it actually matches better the coredump files ? Landry
Re: cvs, stop ignoring *@#%@# files/directories named core
Date: Tue, 20 Mar 2012 10:07:35 + From: Stuart Henderson s...@spacehopper.org core dumps on OpenBSD are not named 'core' anyway so ignoring them is pointless and gets in the way. OK? Why should cvs on OpenBSD behave different from cvs on other platforms? People should not name files or directories core. It leads to accidents like sysadmins accidentally removing stuff when trying to free up disk space by removing all core files. Index: gnu/usr.bin/cvs/src/ignore.c === RCS file: /cvs/src/gnu/usr.bin/cvs/src/ignore.c,v retrieving revision 1.15 diff -u -p -r1.15 ignore.c --- gnu/usr.bin/cvs/src/ignore.c 5 Jan 2012 19:45:20 - 1.15 +++ gnu/usr.bin/cvs/src/ignore.c 20 Mar 2012 10:06:31 - @@ -33,7 +33,7 @@ static int ign_size;/* This many slot static int ign_hold = -1;/* Index where first temporary item * is held */ -const char *ign_default = . .. core RCSLOG tags TAGS RCS SCCS .make.state\ +const char *ign_default = . .. RCSLOG tags TAGS RCS SCCS .make.state\ .*.swp\ .nse_depinfo #* .#* cvslog.* ,* CVS CVS.adm .del-* *.a *.olb *.o *.obj\ *.so *.Z *~ *.old *.elc *.ln *.bak *.BAK *.orig *.rej *.exe _$* *$ *.depend;
Re: cvs, stop ignoring *@#%@# files/directories named core
On 2012/03/20 11:42, Mark Kettenis wrote: Date: Tue, 20 Mar 2012 10:07:35 + From: Stuart Henderson s...@spacehopper.org core dumps on OpenBSD are not named 'core' anyway so ignoring them is pointless and gets in the way. OK? Why should cvs on OpenBSD behave different from cvs on other platforms? That isn't anything new with my diff, it already behaves differently from cvs on other platforms and has done for some time (e.g. *.depend). I can always adjust .cvsrc but I've seen enough other people having problems with these that I think the change is worthwhile. People should not name files or directories core. It leads to accidents like sysadmins accidentally removing stuff when trying to free up disk space by removing all core files. Perhaps, but people do that anyway, and I don't see that as a reason to make it awkward to use them with cvs. It doesn't skip files/dirs with embedded space characters by default either which I think are far more dangerous.
Re: cvs, stop ignoring *@#%@# files/directories named core
On Tue, Mar 20, 2012 at 10:39:02AM +, Stuart Henderson wrote: On 2012/03/20 11:27, Landry Breuil wrote: On Tue, Mar 20, 2012 at 10:07:35AM +, Stuart Henderson wrote: core dumps on OpenBSD are not named 'core' anyway so ignoring them is pointless and gets in the way. Why not making it '*.core' so it actually matches better the coredump files ? Landry Oh yes, that's a better idea. Index: gnu/usr.bin/cvs/src/ignore.c === RCS file: /cvs/src/gnu/usr.bin/cvs/src/ignore.c,v retrieving revision 1.15 diff -u -p -r1.15 ignore.c --- gnu/usr.bin/cvs/src/ignore.c 5 Jan 2012 19:45:20 - 1.15 +++ gnu/usr.bin/cvs/src/ignore.c 20 Mar 2012 10:33:03 - @@ -33,8 +33,8 @@ static int ign_size;/* This many slot static int ign_hold = -1;/* Index where first temporary item * is held */ -const char *ign_default = . .. core RCSLOG tags TAGS RCS SCCS .make.state\ - .*.swp\ +const char *ign_default = . .. RCSLOG tags TAGS RCS SCCS .make.state\ + .*.swp *.core\ .nse_depinfo #* .#* cvslog.* ,* CVS CVS.adm .del-* *.a *.olb *.o *.obj\ *.so *.Z *~ *.old *.elc *.ln *.bak *.BAK *.orig *.rej *.exe _$* *$ *.depend; You have my okay.
Re: cvs, stop ignoring *@#%@# files/directories named core
Why should cvs on OpenBSD behave different from cvs on other platforms? It already does. But largely, lots of people don't use cvs to the extent we do. Of course, any other svn they use will have similar issues. People should not name files or directories core. It leads to accidents like sysadmins accidentally removing stuff when trying to free up disk space by removing all core files. Yes, people shouldn't do that. However, someone has done that, and I suppose the lesson is that people shouldn't delete files or directories called core...