On Friday 02 of July 2010 17:14:09 Kamil Dudka wrote:
> On Friday 02 of July 2010 16:59:31 Kamil Dudka wrote:
> > On Thursday 01 of July 2010 22:39:53 Daniel Stenberg wrote:
> > > On Wed, 30 Jun 2010, Kamil Dudka wrote:
> > > > Attached is my proposal how it may look like. To be honest, I am not
> > > > sure how portable this solution is...
> > >
> > > Wow, I like it. Nice work! The fact that it works even without a debug
> > > build is really great. We could actually make it possible to switch on
> > > this static host name for debug-builds with an environment variable or
> > > so, to allow targets that don't support LD_PRELOAD to also run the
> > > tests.
> > >
> > > And to get the test cases only run on support targets/builds, we have
> > > <precheck> in the test cases that can test for first LD_PRELOAD being
> > > support and as a second option if debug is enabled and only run the
> > > test then.
> >
> > Well, let's go for the second iteration. The only problem I am aware of
> > now is that we override the value of LD_PRELOAD specified by user. I
> > would need to do something like:
> >
> > LD_PRELOAD=./libtest/.libs/libhostname.so:$LD_PRELOAD
> >
> > ... but the perl script does not seem to be ready for this. Any idea?
>
> And please ignore the file 'tests/libtest/hostname.c'. I thought I had
> deleted it, but something went wrong. It will go away.

Oops, and the code shift within runtests.pl also doesn't look perfect as it
fails to skip some test-cases now. I'll probably need some help with the
perl script...

> > Kamil
-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette: http://curl.haxx.se/mail/etiquette.html
Received on 2010-07-02