Downloaded snapshot below - got warning about signature mismatch on
mktmpdir.c (not too surprising).  t/1-basic.t is OK - but that's all that
ran.  Not sure whether myldr/par was tested or not - I think not because of
the test error.  To the extent that the testing is adequate, the fix is
adequate - I didn't get a core dump as I did originally.  But I'm not sure
that the necessary testing was done.

I did rename t/0-signature.t so it would not be run, and make test did a
'cd myldr' and 'perl -e1' and that's all...

--
Jonathan Leffler ([EMAIL PROTECTED])
STSM, Informix Database Engineering, IBM Data Management
4100 Bohannon Drive, Menlo Park, CA 94025
Tel: +1 650-926-6921   Tie-Line: 630-6921
      "I don't suffer from insanity; I enjoy every minute of it!"




|---------+---------------------------->
|         |           Autrijus Tang    |
|         |           <[EMAIL PROTECTED]|
|         |           s.org>           |
|         |                            |
|         |           05/27/2003 09:49 |
|         |           PM               |
|         |                            |
|---------+---------------------------->
  
>---------------------------------------------------------------------------------------------------------------------------------------------|
  |                                                                                    
                                                         |
  |       To:       Nicholas Clark <[EMAIL PROTECTED]>                                 
                                                             |
  |       cc:       Autrijus Tang <[EMAIL PROTECTED]>, Jonathan Leffler/Menlo 
Park/[EMAIL PROTECTED], [EMAIL PROTECTED], [EMAIL PROTECTED]              |
  |       Subject:  Re: PAR 0.68 - core dump on Solaris 7, Perl 5.8.0                  
                                                         |
  |                                                                                    
                                                         |
  
>---------------------------------------------------------------------------------------------------------------------------------------------|




On Tue, May 27, 2003 at 08:53:16PM +0100, Nicholas Clark wrote:
> The SEGV was due to getenv() calling strlen() on its argument, and strlen
> SEGV-ing because the argument (tmpdir) was NULL. I *think* that the logic
> of mktmpdir.c is buggy. (See the patch context)

It looks like that I did an "oops" when cleaning up Markus's code.
The patch looks correct; can people with install problems confirm
that this patch works?  If yes, I'll do a 0.69 asap.

The usual snapshot tarball which contains this change resides at:

    http://p4.elixus.org/snap/PAR.tar.gz

Thanks,
/Autrijus/

> --- myldr/mktmpdir.c.orig   2003-05-25 19:07:50.000000000 +0100
> +++ myldr/mktmpdir.c  2003-05-27 20:37:11.000000000 +0100
> @@ -44,7 +44,7 @@
>
>      for ( i = 0 ; tmpdir == NULL && strlen(tmpval = tmpenv[i]) > 0 ; i++
) {
>          /* fprintf(stderr, "%s: testing env var %s.\n", argv[0],
tmpval); */
> -        if ( (envtmp = getenv(tmpdir)) )
> +        if ( (envtmp = getenv(tmpval)) )
>          {
>              if ( PAR_lstat(envtmp, &statbuf) == 0 &&
>                   ( S_ISDIR(statbuf.st_mode) ||
 >


#### C.DTF has been removed from this note on May 28, 2003 by Jonathan
Leffler


Reply via email to