Re: CVS update [cvs1-11-x-branch]: /ccvs/src/
[email protected] (Larry Jones) Sat, 28 Aug 2004 20:26:25 -0400 (EDT)
| Newsgroups | gmane.comp.lib.gnulib.bugs,gmane.comp.version-control.cvs.bugs |
|---|---|
| Message-ID | <[email protected]> |
Derek Robert Price writes: > > What about the case in question, when output is redirected, piped, and > some other program sets O_NONBLOCK? Ideally, stdio would handle this, > but it currently does not. I've changed my mind, we shouldn't change CVS at all. Nonblocking files do not behave the way files are supposed to, so they should only be used in carefully controlled situations (basically, when you're guaranteed to have exclusive usage of the file). By setting one of its standard files into nonblocking mode, SSH is being extremely antisocial since the file might well be shared with a completely unrelated application which the nonblocking mode will seriously jeopardize (viz CVS, but other applications have the same problem). Thus, the SSH folks should be severely beaten (metaphorically, of course) until they agree to fix their code. I understand what they're trying to accomplish, but they're going about it in the wrong way -- they should be testing for writability with select before trying to write to stderr rather than just blindly putting stderr into nonblocking mode and expecting everyone else to just cope with it. -Larry Jones We don't ATTEND parties, we just CRASH 'em. -- Calvin