<p><br /> On Dec 17, 2010 8:02 AM, "Craig Ringer" <<a
href="mailto:craig@postnewspapers.com.au">craig@postnewspapers.com.au</a>>wrote:<br /> ><br /> > On 16/12/10
21:01,Magnus Hagander wrote:<br /> ><br /> > > Found another problem in it: when running with an older version
of<br/> > > dbghelp.dll (which I was), it simply didn't work. We need to grab the<br /> > > version of
dbghelp.dllat runtime and pick which things we're going to<br /> > > dump based on that.<br /> ><br /> > I
wasabout to suggest dropping the fallback loading of the system<br /> > dbghelp.dll, and just bundle a suitable
versionwith Pg, then I saw how<br /> > big the DLL is. It's 800kb (!!) of code that'll hopefully never get used<br
/>> on any given system. With a footprint like that, bundling it seems<br /> > rather less attractive.<br />
><br/> > I think Magnus is right: test the dbghelp.dll version and fall back to<br /> > supported features -
asfar back as XP, anyway; who cares about anything<br /> > earlier than that. An updated version can always be put
inplace by the<br /> > user if the built-in one can't produce enough detail.<p>Did you get a chance to test that it
stillproduced a full dump on your system? <p>> > We use the existance of the "crashdumps" directory as an
indicationwe<br /> > > want crashdumps. That's fine when the system is up. But what if we<br /> > > crash
*inthe postmaster before we have done chdir()*?<br /> > ><br /> > > Should we perhaps instead define a
subdirectoryof *where the .EXE<br /> > > file is*, and dump the file there?<p>... <p>> Is this going to be a
real-worldissue?<p>Nah, I'm with tom on the fact that this is probably not. <br /><p>> > It needs to be
documentedsomewhere.Perhaps in "15.8 Platform Specific<br /> > > Notes"? That's really about building it, but it
mightbe reasonable<br /> > > there anyway? It does hold a number of things today that aren't<br /> > >
relatedto building, for other platforms.<br /> ><br /> > Seems reasonable.<p>I'll put something together there
then.<p>/Magnus