PostgreSQL and CVS DLL

Christopher Faylor
Wed Jan 9 11:10:00 GMT 2002

On Wed, Jan 09, 2002 at 01:57:10PM -0500, Norman Vine wrote:
>Christopher Faylor writes:
>>On Wed, Jan 09, 2002 at 12:01:21PM -0500, Jason Tishler wrote:
>>>On Wed, Jan 09, 2002 at 11:39:56AM -0500, Norman Vine wrote:
>>>> OK -- using snapshots from
>>>> postgres runs fine using snapshot DLL dated 20011231
>>>> crashes for me with all DLL's dated 20020104 and later
>>>> I have a locally built DLL 20020103 that also crashes
>>>Please try to determine what specific Cygwin change is causing the
>>>postmaster problem.  Since the date range is small, searching
>>>cygwin-cvs should not take too long.
>>Where is the analysis of *where* the program died?  That should be
>>pretty simple to show.  Look at how-to-debug-cygwin.txt for details.
>I realized that this was a 'crappy bug report' and even mentioned so in
>my post but I thought, perhaps wrongly, that this was worth a 'heads
>up' as I already had several hours into debugging this to no avail with
>fetching the snapshot dll's, compiling both postgres and cygwin with
>and without -g running the regression test several times ect.., and am
>not sure when my next couple of free hours are going to be, in order to
>dig into this deeper,

I think I've pointed out that the *majority* of the bug reports here are
"crappy", right?

I am trying to reorient your (and everyone's) thinking.  You spent a
long time debugging.  That is appreciated.  However, you are a member of
the mailing list called 'cygwin-developers'.  That means that you should
have committed, at some point, to be contributing to cygwin.  I usually
ask people if they have built cygwin when I grant them the awe inspiring
membership in this mailing list.

The answer has to be "yes" in order to gain membership.  Maybe I didn't
ask the question of everyone.  If not, then I guess you're off the hook.
If you did commit to having built cygwin then I think I have a right
to assume that you should be able to generate a debugging version of
cygwin and provide an analysis of any stackdump file.  That shouldn't
be that big a deal.  It takes an extra five minutes to figure out
exactly where the program crashed from the stackdump file.

If you are not set up to do this or don't want to do this then send "it
crashes" bug reports to  I don't want to get into the
habit of bug reports with no analysis here.  That is not the intent of
this mailing list.


More information about the Cygwin-developers mailing list