Donate Share

Radmind

Code

Programming Languages: C

License: BSD License

Repositories

browse code, statistics, last commit on 2009-10-16 cvs -d:pserver:anonymous@radmind.cvs.sourceforge.net:/cvsroot/radmind login

cvs -z3 -d:pserver:anonymous@radmind.cvs.sourceforge.net:/cvsroot/radmind co -P modulename

browse code, statistics git://radmind.git.sourceforge.net/gitroot/radmind/radmind (read-only)

Show:

What's happening?

  • Comment: -p option for ra.sh (for port)

    Committed.

    2009-11-11 18:29:13 UTC by fitterhappier

  • Comment: Add a copy mode to lmerge

    Committed with a few very minor tweaks. Thanks for your contribution.

    2009-11-11 16:45:54 UTC by fitterhappier

  • Comment: Add a copy mode to lmerge

    I've been making changes based of 1.13.0 but it does apply against the git head with just an offset.

    2009-11-11 00:49:02 UTC by abic

  • Comment: Add a copy mode to lmerge

    Much better. Just a few more comments. Get these things corrected, and I'll commit the patch. 1) Initialize rc to -1 so you don't have to set rc = -1 for each error case. The only place you'll be setting rc = 0 is after the copy has succeeded, just before the "cleanup" goto target. 2) We don't want to retain modtimes. The copy feature is desirable because the admin wants an explicit break...

    2009-11-10 21:37:05 UTC by fitterhappier

  • Comment: Add a copy mode to lmerge

    Still not sure if I'm handling errors as cleanly as I could but a 3rd patch is up.

    2009-11-10 19:00:48 UTC by abic

  • Comment: Add a copy mode to lmerge

    ENOENT would only be returned by a failed open of dest when the errno was ENOENT. Other errors would return -1. mkdirs is effectively mkdir -p. We're trying to avoid multiple stat+mkdir calls back through the path elements by detecting ENOENT when we attempt to create the destination file. Rather than stat'ing, e.g., file/loadset.T/opt/bin/apache/modules, getting ENOENT, then stat'ing...

    2009-11-09 18:03:28 UTC by fitterhappier

  • Comment: Add a copy mode to lmerge

    It seems that only the errno from open dest_file would be a case where mkdirs would resolve the issue, so ENOENT from the open src_file or from utimes would then also cause mkdirs to run if I'm just returning the error code to the caller. Similarly if opath didn't exist when link was ran you'd also have a case where mkdirs is getting run. Thoughts on this? Would it be better to do a stat or...

    2009-11-05 19:27:39 UTC by abic

  • Comment: Add a copy mode to lmerge

    Man, I wish SF.net preserved my formatting. Hopefully it's OK in the e-mail notification.

    2009-11-03 19:25:15 UTC by fitterhappier

  • Comment: Add a copy mode to lmerge

    The link error handling isn't ideal. It should probably be checking to see if errno is ENOENT before attempting to create the directory structure for the path. Your code should do something similar, maybe like this: if (( destfd = open( dest, O_WRONLY, mode )) < 0 ) { if ( errno == ENOENT ) { rc = errno; goto cleanup; /* close fds, etc. */ } rc = -1;...

    2009-11-03 19:24:42 UTC by fitterhappier

  • Comment: fsdiff: negative directories with xattrs

    Duplicate of 2887960.

    2009-10-30 18:46:08 UTC by fitterhappier

Our Numbers