hi,
>> - someone asked me for mod_vhost_alias on Win32, so I compiled it
>> and it loads into Server, but didnt test more with it; can we add
>> this to the 1.3 Win32 project or does someone think it will not run
>> for some reason? (I didnt look at the code because it compiled
>> without problems).

> Agreed - I'll commit a project file for this.
ok, thanks.

>> - as there is a makefile for mod_example for unix present also there
>> should be one for other platforms as Win32 and NetWare too, not build
>> by default but present in the project.

> No.  APXS works for win32.  I'd be happy to document how to build
> mod_example by APXS for users to follow.
hm, ok; but then you should remove the makefile for Unix because APXS works also for 
Unix.

>> - the manual doesnt work after build, at least not on Win32 and NetWare.
>> I have to add always the 'Includes' and 'AddHandler server-parsed .html'
>> directives. As you know I posted this issue already to the list and
>> someone
>> told me that the files will be expanded; but at least with NetWare I'm
>> 100%
>> sure that there expands nothing and on Win32 also after 'InstallBin' the
>> manual isnt functional. I think it would be nice if we add these two
>> directives to the new directory block of ./manual because then we have
>> immediatly after a build also a functioning sample for SSI:

> I disagree.  Please see the expand.pl and notes in 'rolling the release
> tarball'
> in the dev.apache.org site.  Users from CVS suffer this same issue.  Users
> of
> the tarball/zipfile do not suffer this problem.

> But the way 1.3 kept the src tree within the build tree is a nightmare.
> We
> can't auto-expand without busting the image so it can no longer be updated
> or
> committed to cvs.  It's just a tangled mess that is resolved in 2.0
hmm, without having looked yet at 'rolling the release tarball' (but will do!) I cant 
understand why we need a perlscript for expanding (Perl is not on every building 
platform!) when we could have a working manual with only one additional and one 
changed line in httpd.conf on ALL platforms; or do you not trust SSI? What is the 
reason against this idea?

>> - currently we have 45 warnings with the Win32 build of 1.3: does it make
>> sense that I look through it and make some casts to make VC6 happy and
>> contribute patches?

> It's a thought.  I'd rather chase the problems out of 2.0 than 1.3 though.
> I cleaned up a ton a while back, and there are many more, but many (most?)
> can't be fixed without substantial changes to code that -can't- be
> changed.
> Win32 uses atypical definitions for things in sockaddr and so on that we
> can't work around portably without introducing -new- compiler warnings on
> other platforms.
ok, that's what I thought, was just to confirm that.

> OTOH, if you find a warning that indicates a real problem, PLEASE bring
> that
> to our attention, and even suggest a patch :)
ok.

>> - I've used a patch against mod_autoindex which enables setting of colors
>> with server generated indexes and it works fine; no I wanted to build
>> mod_autoindex as separate module, but when I do so mod_autoindex runs
>> fine
>> but 'MultiViews' doesnt work anymore! Is this because of wrong load
>> order?
>> I tested this only with NetWare, so it may be an issue with this
>> platform...

> -1 for these sorts of changes to core for the 1.3 tree - but to answer to
no, I didnt want to get these changes in! That was the reason why I tried buidling 
mod_autoindex as separate module.

> your
> question, how doesn't it work in multiviews?  Please change the subject
> and
> split this question from the other (1.3 release-releated) topics.
ok. Will do so but first test on other platforms and see if there is same bahavior...

> In general, please consider bandaids for 1.3/Win32 very temporary, once
> 2.0
> is released as GA quality, we will quickly orphan Win32, in the sense that
> the
> 1.3 implementation is far too hackish for ongoing support ;)  You can't
> have a
> clean 1.3 implementation, because NT didn't fit in the 1.3 schema.  All
> that
> code is along it's own fork, with all sorts of messy thunking.  APR
> resolves
> the 80/20 problem.
ok, but for now we have only this...

thanks for detailed answers.

Guenter.

Reply via email to