[pacman-dev] Fwd: CVS update of pacman-lib/lib/libalpm (remove.c)
I just wanted to run this by anyone interested. Below is what I've been using for CVS commits, sent only to myself for the time being. Is including the full diff output a good idea, or over the top? Let me know.. ---------- Forwarded message ---------- From: Aaron Griffin <aaron@archlinux.org> Date: Jan 18, 2007 11:04 AM Subject: CVS update of pacman-lib/lib/libalpm (remove.c) To: aaronmgriffin@gmail.com Date: Thursday, January 18, 2007 @ 12:04:25 Author: aaron Path: /home/cvs-pacman/pacman-lib/lib/libalpm Modified: remove.c (1.54 -> 1.55) Some confusion with variable names in unlink_file caused unlink to try and find the _relative_ file instead of absolute. The absolute path name has also been used in some of the output there too. ----------+ remove.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-) Index: pacman-lib/lib/libalpm/remove.c diff -u pacman-lib/lib/libalpm/remove.c:1.54 pacman-lib/lib/libalpm/remove.c:1.55 --- pacman-lib/lib/libalpm/remove.c:1.54 Thu Jan 18 11:04:05 2007 +++ pacman-lib/lib/libalpm/remove.c Thu Jan 18 12:04:25 2007 @@ -168,7 +168,7 @@ double percent = 0.0; char *file = lp->data; char line[PATH_MAX+1]; - char *checksum = _alpm_needbackup(file, info->backup); + char *checksum = _alpm_needbackup(lp->data, info->backup); if ( *position != 0 ) { percent = (double)*position / filenum; @@ -183,15 +183,15 @@ } snprintf(line, PATH_MAX, "%s%s", handle->root, file); if ( lstat(line, &buf) ) { - _alpm_log(PM_LOG_DEBUG, _("file %s does not exist"), file); + _alpm_log(PM_LOG_DEBUG, _("file %s does not exist"), line); return; } if ( S_ISDIR(buf.st_mode) ) { if ( rmdir(line) ) { /* this is okay, other pakcages are probably using it (like /usr) */ - _alpm_log(PM_LOG_DEBUG, _("keeping directory %s"), file); + _alpm_log(PM_LOG_DEBUG, _("keeping directory %s"), line); } else { - _alpm_log(PM_LOG_DEBUG, _("removing directory %s"), file); + _alpm_log(PM_LOG_DEBUG, _("removing directory %s"), line); } } else { /* check the "skip list" before removing the file. @@ -206,7 +206,7 @@ } if ( skipit ) { _alpm_log(PM_LOG_FLOW2, _("skipping removal of %s as it has moved to another package"), - file); + line); } else { /* if the file is flagged, back it up to .pacsave */ if ( nb ) { @@ -217,16 +217,16 @@ char newpath[PATH_MAX]; snprintf(newpath, PATH_MAX, "%s.pacsave", line); rename(line, newpath); - _alpm_log(PM_LOG_WARNING, _("%s saved as %s"), file); + _alpm_log(PM_LOG_WARNING, _("%s saved as %s"), line, newpath); } } } else { - _alpm_log(PM_LOG_FLOW2, _("unlinking %s"), file); + _alpm_log(PM_LOG_FLOW2, _("unlinking %s"), line); int list_count = _alpm_list_count(trans->packages); /* this way we don't have to call _alpm_list_count twice during PROGRESS */ PROGRESS(trans, PM_TRANS_PROGRESS_REMOVE_START, info->name, (double)(percent * 100), list_count, (list_count - _alpm_list_count(targ) + 1)); ++(*position); } - if (unlink(file) == -1) { + if (unlink(line) == -1) { _alpm_log(PM_LOG_ERROR, _("cannot remove file %s: %s"), file, strerror(errno)); } }
2007/1/18, Aaron Griffin <aaronmgriffin@gmail.com>:
I just wanted to run this by anyone interested. Below is what I've been using for CVS commits, sent only to myself for the time being.
Is including the full diff output a good idea, or over the top? Let me know..
Looks good for me (except that Gmail message text displaying sucks). I used for diffs from sdl-cvs ML. -- Roman Kyrylych (Роман Кирилич)
On 1/18/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
Looks good for me (except that Gmail message text displaying sucks). I used for diffs from sdl-cvs ML.
Can you explain? I'm not on that ML. There are a few scripts which do fancy colored diffs, but that requires HTML email and we all know plain text is superior :) I've come across alot of text-wrapping issues with gmail lately... I need to figure out if something can be done about that, or scrap gmail altogether....
2007/1/18, Aaron Griffin <aaronmgriffin@gmail.com>:
On 1/18/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
Looks good for me (except that Gmail message text displaying sucks). I used for diffs from sdl-cvs ML.
Can you explain? I'm not on that ML. There are a few scripts which do fancy colored diffs, but that requires HTML email and we all know plain text is superior :)
HTML messages are evil. >:-E
I've come across alot of text-wrapping issues with gmail lately... I need to figure out if something can be done about that, or scrap gmail altogether....
I saw somewhere a nice set of CSS themes for Gmail that make it look nicer, but don't remember now where there were. I think I'll made at least fixed-width mod when I'll get very angry on gmail. Dunno how to deal with those stupid wraps yet. Here's the a part of a message from sdl-cvs ML: Revision: 2911 Author: pmandin Date: 2006-11-13 12:34:18 -0800 (Mon, 13 Nov 2006) Log Message: ----------- Save/restore current video mode and palette Modified Paths: -------------- trunk/SDL/src/video/xbios/SDL_xbios.c trunk/SDL/src/video/xbios/SDL_xbios.h trunk/SDL/src/video/xbios/SDL_xbiosmodes.c Added Paths: ----------- trunk/SDL/src/video/xbios/SDL_xbiosmodes.h Modified: trunk/SDL/src/video/xbios/SDL_xbios.c =================================================================== --- trunk/SDL/src/video/xbios/SDL_xbios.c 2006-11-13 19:59:28 UTC (rev 2910) +++ trunk/SDL/src/video/xbios/SDL_xbios.c 2006-11-13 20:34:18 UTC (rev 2911) @@ -174,7 +174,7 @@ { /* Save screensaver settings */ - /* Init video mode list */ + /* Init video mode list, save current video mode settings */ SDL_XBIOS_InitModes(_this); return (0); @@ -187,5 +187,6 @@ /* Restore screensaver settings */ + /* Restore previous video mode settings */ SDL_XBIOS_QuitModes(_this); } Modified: trunk/SDL/src/video/xbios/SDL_xbios.h =================================================================== --- trunk/SDL/src/video/xbios/SDL_xbios.h 2006-11-13 19:59:28 UTC (rev 2910) +++ trunk/SDL/src/video/xbios/SDL_xbios.h 2006-11-13 20:34:18 UTC (rev 2911) @@ -29,8 +29,13 @@ . . . skipped . . . Looks slightly better for me. Though, as you may notice - this is generated from SVN (they swithched to SVN before starting real work on SDL 1.3). The only improvement I would like to see in pacman cvs reports - make this:
----------+ remove.c | 16 ++++++++-------- 1 file changed, 8 insertions(+), 8 deletions(-)
Index: pacman-lib/lib/libalpm/remove.c diff -u pacman-lib/lib/libalpm/remove.c:1.54 pacman-lib/lib/libalpm/remove.c:1.55 --- pacman-lib/lib/libalpm/remove.c:1.54 Thu Jan 18 11:04:05 2007 +++ pacman-lib/lib/libalpm/remove.c Thu Jan 18 12:04:25 2007
look like just
--- pacman-lib/lib/libalpm/remove.c:1.54 Thu Jan 18 11:04:05 2007 +++ pacman-lib/lib/libalpm/remove.c Thu Jan 18 12:04:25 2007
because there is
Path: /home/cvs-pacman/pacman-lib/lib/libalpm
Modified: remove.c (1.54 -> 1.55)
already. But that's just cosmetics. :-) -- Roman Kyrylych (Роман Кирилич)
On 1/18/07, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
I just wanted to run this by anyone interested. Below is what I've been using for CVS commits, sent only to myself for the time being.
Is including the full diff output a good idea, or over the top? Let me know..
I wish we were on something other than CVS, because then you could link to a URL of the patchset. CVS doesn't do that very well, obviously, as it is a per-file revision system. It isn't like I have a problem with mailbox space using Gmail, so I don't care either way about full diffs. -Dan
2007/1/18, Dan McGee <dpmcgee@gmail.com>:
On 1/18/07, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
I just wanted to run this by anyone interested. Below is what I've been using for CVS commits, sent only to myself for the time being.
Is including the full diff output a good idea, or over the top? Let me know..
I wish we were on something other than CVS, because then you could link to a URL of the patchset. CVS doesn't do that very well, obviously, as it is a per-file revision system.
It isn't like I have a problem with mailbox space using Gmail, so I don't care either way about full diffs.
Heh, does gmail is used by 99% people now? :-D SVN would be nice, but there's no big need to switch now, IMHO. -- Roman Kyrylych (Роман Кирилич)
On 1/18/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
2007/1/18, Dan McGee <dpmcgee@gmail.com>:
I wish we were on something other than CVS, because then you could link to a URL of the patchset. CVS doesn't do that very well, obviously, as it is a per-file revision system.
SVN would be nice, but there's no big need to switch now, IMHO.
Yeah, I actually like that CVS is file based instead of changeset based, but that's personal preference. This might be a good opportunity to try out monotone[1], which I'd been interested in for some time. It's similar to git in some ways. Here's a "ViewMTN" web interface: http://viewmtn.angrygoats.net/ 1: http://venge.net/monotone/
On 1/18/07, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On 1/18/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
2007/1/18, Dan McGee <dpmcgee@gmail.com>:
I wish we were on something other than CVS, because then you could link to a URL of the patchset. CVS doesn't do that very well, obviously, as it is a per-file revision system.
Another note on the topic: CVS can "maintain" changesets in one commit (the mail script does it) it just requires some extra work to catalog all the files that were added in a given commit. cvstrac[1] does this for it's web interface: http://www.cvstrac.org/cvstrac/about
On 1/18/07, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On 1/18/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
2007/1/18, Dan McGee <dpmcgee@gmail.com>:
I wish we were on something other than CVS, because then you could link to a URL of the patchset. CVS doesn't do that very well, obviously, as it is a per-file revision system.
SVN would be nice, but there's no big need to switch now, IMHO.
Yeah, I actually like that CVS is file based instead of changeset based, but that's personal preference.
This might be a good opportunity to try out monotone[1], which I'd been interested in for some time. It's similar to git in some ways. Here's a "ViewMTN" web interface: http://viewmtn.angrygoats.net/
On the subject of possible alternatives, bazaar is quite nice ( http://bazaar-vcs.org ), and it has a web-interface ( https://launchpad.net/bzr-webserve if you click on the link to its home page, you can see what the interface looks like 'cause the site itself uses itself (if that makes sense :P) ). ~ Jamie / yankees26
2007/1/18, James Rosten <seinfeld90@gmail.com>:
On 1/18/07, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On 1/18/07, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
2007/1/18, Dan McGee <dpmcgee@gmail.com>:
I wish we were on something other than CVS, because then you could link to a URL of the patchset. CVS doesn't do that very well, obviously, as it is a per-file revision system.
SVN would be nice, but there's no big need to switch now, IMHO.
Yeah, I actually like that CVS is file based instead of changeset based, but that's personal preference.
This might be a good opportunity to try out monotone[1], which I'd been interested in for some time. It's similar to git in some ways. Here's a "ViewMTN" web interface: http://viewmtn.angrygoats.net/
On the subject of possible alternatives, bazaar is quite nice ( http://bazaar-vcs.org ), and it has a web-interface ( https://launchpad.net/bzr-webserve if you click on the link to its home page, you can see what the interface looks like 'cause the site itself uses itself (if that makes sense :P) ).
Anyway, I would like to see cvs commits autoposted here soon. ;-) -- Roman Kyrylych (Роман Кирилич)
participants (4)
-
Aaron Griffin
-
Dan McGee
-
James Rosten
-
Roman Kyrylych