|
|
Boost Testing : |
From: Robert Ramey (ramey_at_[hidden])
Date: 2004-10-19 19:20:47
"Markus Schöpflin" <markus.schoepflin_at_[hidden]> wrote in message
news:cl3q15$bji$1_at_sea.gmane.org...
> Robert Ramey wrote:
>
> > Markus,
> >
> > Thanks for testing on your 64 bit systems.
>
> You're welcome. ;-)
>
> > The serialization library is passing 100% with gcc on your platform.
>
> Yes, I have noted this, it's great. But I have to admit I'm not 100%
> happy that you didn't choose the approach proposed by John on the 64 bit
> overloads. And there are some warning issued by GCC when compiling the
> library, but this has been noted on the boost mailing list already.
It wasn't a matter of choice. I couldn't make it work for my case. I've
made a couple of adjustments which I believe will make those warnings go
away.
> > The other compiler ( compac / alpa tru64 ?) is failing compile one code
> > module. This is src/xml_grammar.cpp line 28
> > This code is a work around from Rogue Wave libray issue. I discovered
and
> > "fixed" an issue while porting to BCB version 5.5.1 . I had hoped that
this
> > "fix" would also apply to previous versions of the RogueWave library .
> > Would it be possible for you to take a look at that. It would be my
hope
> > that with some sort of adjustment here, the libraries would build and
tests
> > would run. I don't know if this is the only issue but its currently
> > stopping the show. I don't know enough about the rogue wave library usd
and
> > have no access it here. Maybe you've had occasion to address simlar
> > situations with other boost libraries and you might have some special
> > insight that reveals the "magic fix".
>
> Just today I comment out this workaround locally to see if I could do
> anything about it. The library compiles ok if the workaround is removed
> but I had unresolved symbols when linking, so I didn't pursue the issue
> any further. If you are interested, I will try to get a small isolated
> test case from the serialization library up and running.
I'm sure that this is why I put in that special code for rogue wave. I
would like to see the list of undefined symbols. I suspect this can be
addressed by using a special workaround for this platform just as I made one
for the one we did for BCB 5.5.1 . I'm inclined to comment out this work
around for this platform and see which tests fail to link in the nightly
test
>
> Maybe we can make the library work for this platform with joint efforts.
>
I would be very pleased to see this work.
Robert Ramey
> Markus