libtool-patches
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: Libtool HEAD on Windows.


From: Duft Markus
Subject: RE: Libtool HEAD on Windows.
Date: Mon, 3 Mar 2008 10:41:36 +0100

Hi!

Ralf Wildenhues <mailto:address@hidden> wrote:
> * Duft Markus wrote on Mon, Mar 03, 2008 at 09:33:44AM CET:
>>>> 
[[...]]
>> 
>> Still waiting for _something_...
> 
> Sorry about being completely silent.  Your patch is intrusive in that
> it changes more than should be necessary, it changes code that should
> not be changed.  The ideal is: only libtool.m4 and ltdl.m4 contain
> system specifics, all other code is generic.  Now I know that we're
> far from this ideal, but every step further away is one that should be
> scrutinized.

Then Windows won't work, i think. I really need to add dllimport stuff
(which is ifdef'd out in first place, and which is there allready
partially for cygwin/mingw...) to c/h files. I tried to change as little
as possible anyway allready :(

I could throw away the changes to the loadlibrary loader, since the
dlopen interface which i implemented is used anyway.
I don't know if it would have any side effects on other software if
RTDL_GLOBAL is not defined. If not, then maybe i could remove the
definition for it to avoid the other ifdef.
As for beeing generic: all the locations where there is a switch/case on
the host, i have to insert winnt.

> 
> Also, some of the testsuite changes look like they are papering over
> the failure rather than fix the issue.  The template.at changes fall
> into this category.

Hmm.... Here i had to add libraries to the command line for the linker
to find symbols. If this isn't desired i'll have to xfail or skip the
test, since i cannot link with unresolved symbols on windows.

> 
> Last but not least, I don't know if you noticed that Libtool was in
> regression fixes mode up to the 2.2 release (yeah, I know it was an
> ugly long time).  Given limited time, I for one simply ignored
> everything that wasn't fixing a regression or easy to verify as safe.

No, i didn't notice that... 

> 
> I might give a detailed review on your patch, but that may be some
> days. 

Thats absolutely no problem, now that i have some status :)

Thanks for the (now) fast answer :)

Cheers, Markus

> 
> Cheers,
> Ralf





reply via email to

[Prev in Thread] Current Thread [Next in Thread]