bug-cvs
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: cvs patches & notes


From: Russ Tremain
Subject: Re: cvs patches & notes
Date: Wed, 9 May 2001 20:58:47 -0700

Hi Rob -

In regards to your logmsg.c patch, I have been able
to kludge around the white space problem by use
%{vsV} in the format specifier, and then joining
the arguments passed by the shell back together
in my loginfo script.  I can then recognize
the white space because it is trapped between
the two revision strings.  Ugly hack though.

Another thought would be to modifiy logmsg.c
to use url-style encoding of the file name.

This would solve the problem once and for all.
Could make it a config option in order to make it backward
compatible.

What are your thoughts?  I'm cc'ing Greg Klanderman
on this because he has also done some recent work in this
area.

-Russ 

(russ.tremain@sun.com)

>From: Rob Saccoccio 
>Subject: cvs patches & notes 
>Date: Fri, 13 Apr 2001 10:03:52 -0700 
>
>
>       I've identified and fixed (or worked around) some issues with cvs and
>       thought I'd pass 'em along...
>
>       1. There's a "bogus kludge" in server.c wrapped by "#ifdef sun" that 
> applies
>       to pretty old unpatched versions of SunOS.  Maybe an autoconf script 
> could
>       be written to test for the condition that causes the problem.  I had a 
> look
>       at when the OS patch was first released (and unfortunately don't recall 
> the
>       details), but it was years ago.  A reasonable alternative may be to 
> assume
>       the kludge isn't needed and suggest that if the problem arises the 
> patch be
>       applied or SUNOS_KLUDGE be defined to enable the old behavior.
>
>       2. When operating in pserver mode, the loginfo scripts are invoked with 
> a
>       file descriptor still open (10) which enables writing back to the 
> client.
>       This is clearly an oversight.  I came across this as I was trying to 
> get the
>       server to return without waiting for the loginfo scripts to complete.  
> In
>       turns out, the server is waiting on the closure of a FD (9) that is 
> closed
>       in the helper client, but was passed to its children.  I have a detailed
>       truss if you'd like it.  I couldn't figure out the "right" place to 
> close
>       the FD, so I now close it at the top of my scripts.
>
>       3. Below is a patch against 1.11 logmsg.c that allows loginfo scripts to
>       properly handle filenames with spaces.  Currently, the arguments are not
>       properly setup to handle this.  The patch is backward compatible.
>
>       --robs
>





reply via email to

[Prev in Thread] Current Thread [Next in Thread]