Re: Darwin patches?

[Max Horn]

At 17:39 Uhr -0500 04.10.2001, scott hutinger wrote:
> On Fri, 5 Oct 2001, Max Horn wrote:
>
> >  >Ok, cdemo tests fail.  What are the requirements from admin for
> >  >modification to cdemo if needed?  Sorry I have to ask, sometimes people
> >  >have various reasons for not allowing specific modifications.
> >
> >  The problem is not in cdemo, this is a test, and it fails for valid
> >  reasons on Darwin - namely, it is the zsh quoting issue strikin once
> >  more. I reported this before, but I am not the one to fix this in
> >  libtool's source - libtool's source ain't easy to understand, yet to
> >  modify in a manner that doesn't break many other systems.
>
> Yes, I was asking what the requirements are for regression testing
> modifications if need be.  Normally this isn't a good thing, and causes
> problems all around for some time (new branch until working again).  But
> in some instances the modifications are small enough not to pull a domino
> effect.  I think everyone knows that anyway.

I understood you were asking for a change to a regression test. And I 
think I explained why this would *not* be a good thing! There is an 
actual problem in libtool, caused by zsh on Darwin. Changing the 
regression test will not fix the problem, only make the regression 
test useless. after all, regression tests are there to flag such 
problems.

So: leave cdemo alone. Rather, try to fix the *source* for the 
problem. Citing an older mail:


At 1:35 Uhr +0200 22.09.2001, Max Horn wrote:
> 1) with regard to dlpreopen, there is yet another quoting problem. 
> The libtool file generated from HEAD-cvs contains three lines like 
> this:
>
> global_symbol_to_c_name_address="sed -n -e 's/^: \([^ ]*\) \$/ 
> {\\"\1\\", (lt_ptr) 0},/p' -e 's/^[BCDEGRST] \([^ ]*\) \([^ ]*\)\$/ 
> {\"\2\", (lt_ptr) \&\2},/p'"


And that is exaclty why cdemo fails.


Max
--
-----------------------------------------------
Max Horn
Software Developer

email: <mailto:address@hidden>
phone: (+49) 6151-494890