Re: [jamboost] Re: regression test for bin.cygwinx86q

From: Vladimir Prus (ghost_at_[hidden])
Date: 2003-04-02 07:47:06

Hi Andre,

> Also, symlink.py is failing because the created links are called
> hello_debug.exe link/hello_release.exe (note the .exe suffix)

Ehm... I'm not sure what is the right solution. Now, 'symlink' retains
target suffices. Is this right idea?

> > 1. Why going through shell is bad?
>
> Because you have to escape everything properly. Tests were failing
> because of an unescaped semicolon that was breaking the command in two.
> In my system, I was having problems also with whitespaces in directory
> names (I was tricked by win xp into using spaces in my username)

Ok, I understand now.

> > 2. Could you please resend the patch as unified diff ("cvs diff -u").
> > Somehow, emacs fails to convert your context diff into unified one, which
> > is much more readable, IMO.
>
> See attached.

Thanks! I've applied the patch with the following additional change:

--- BoostBuild.py.prev	Wed Apr 2 16:23:14 2003
+++ BoostBuild.py	Wed Apr 2 16:26:10 2003
@@ -224,10 +224,10 @@
try:
kw['program'] = []
kw['program'] += self.program
+ if extra_args:
+ kw['program'] += extra_args.split(" ") 
if pass_toolset:
kw['program'].append(self.toolset)
- if extra_args:
- kw['program'] += extra_args.split(" ")
kw['chdir'] = subdir
apply(TestCmd.TestCmd.run, [self], kw)
except:

The same change was made in revision 1.47 of BoostBuild.py and I'm redoing it 
for the same reason -- because passing "gcc -n" to bjam results in "-n" being 
ignored.

Thanks,
Volodya