You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by Justin Erenkrantz <je...@ebuilt.com> on 2001/07/30 00:59:14 UTC
Re: cvs commit: httpd-2.0/server test_char.h
On Sun, Jul 29, 2001 at 10:27:30PM -0000, jwoolley@apache.org wrote:
> jwoolley 01/07/29 15:27:30
>
> Removed: server test_char.h
> Log:
> Nope, changed my mind. With this hack, the build only works right the first
> time... a make clean will rm test_char.h. If Solaris 7 is broken, it's broken.
>
> There's got to be a better way. But not for 2.0.22.
If it's in CVS, then when you do a commit, you get the generated
test_char.h. Not what we want...
I think we could just add a blank copy to the tarball for 2.0.22 (i.e.
not to CVS). Jeff or someone who cares about supporting system make on
Solaris 7 can come up with some clever way, or we just always add
a blank copy of that file to the tarball.
The second time they do a make with Solaris 7, it'll build fine because
test_char.h will now exist. That'd be the obvious thing to do if you
are faced with "this doesn't work" - try again. In this case, it'd
work doing it a second time... -- justin
Re: cvs commit: httpd-2.0/server test_char.h
Posted by Justin Erenkrantz <je...@ebuilt.com>.
On Sun, Jul 29, 2001 at 07:14:09PM -0400, Cliff Woolley wrote:
> Still, something rubs me the wrong way about the
> discrepancy.
Yup. This looks like a legitimate bug in Solaris 7. But, it isn't
present in Solaris 8, so someone at Sun agrees this behavior is bogus.
(My Solaris 7 box has the latest recommended patches from Sun, so it
should be up-to-date.)
Jeff may have some thoughts whenever he reads my messages. I'm not sure
if that was what he was seeing on Solaris - it definitely expanded the
wildcards for me, but it blew up on the lack of test_char.h.
I bet there is some way to work around this. I just don't have the time
and inclination to research this further. =-) -- justin
Re: cvs commit: httpd-2.0/server test_char.h
Posted by Justin Erenkrantz <je...@ebuilt.com>.
On Sun, Jul 29, 2001 at 04:06:14PM -0700, Ryan Bloom wrote:
> Guys, test_char.h was used in apache-1.3 as well as 2.0, and we didn't have this proble
> with Solaris 7 and apache 1.3. Has anybody looked to see what the difference is?
If I take out the include directives (i.e. manually expand everything)
and switch the all target to be:
all: test_char.h all-recursive
it works (this is close to what Apache 1.3 has).
But, if we switch local-all to be:
local-all: test_char.h $(TARGETS)
the dependency check on test_char.h in util.lo is still screwed up
(or if you let it be expanded by the LTLIBRARY_SOURCES value - which
is how it is now in CVS).
It looks like the dependencies aren't read correctly in Solaris 7 when
we execute $(MAKE) local-all from the all-recursive target. We don't
use $(MAKE) in Apache 1.3 from what I can tell - I think I've heard
reports of $(MAKE) being busted in Solaris's make (but I'm not sure).
Does that help? I'm at the limits of make knowledge here.
I'm supposed to be writing a design document right now. -- justin
Re: cvs commit: httpd-2.0/server test_char.h
Posted by Ryan Bloom <rb...@covalent.net>.
> > I think we could just add a blank copy to the tarball for 2.0.22 (i.e.
> > not to CVS). Jeff or someone who cares about supporting system make on
> > Solaris 7 can come up with some clever way, or we just always add
> > a blank copy of that file to the tarball.
>
> Done, begrudgingly. :-/ We have to come up with a more permanent
> solution--as it is, if you check out APACHE_2_0_22 from CVS, you don't get
> the file, but if you unzip the tarball, you do [granted, it's a zero-byte
> file at that point]. Still, something rubs me the wrong way about the
> discrepancy.
Guys, test_char.h was used in apache-1.3 as well as 2.0, and we didn't have this proble
with Solaris 7 and apache 1.3. Has anybody looked to see what the difference is?
Ryan
_____________________________________________________________________________
Ryan Bloom rbb@apache.org
Covalent Technologies rbb@covalent.net
-----------------------------------------------------------------------------
Re: cvs commit: httpd-2.0/server test_char.h
Posted by Cliff Woolley <cl...@yahoo.com>.
On Sun, 29 Jul 2001, Justin Erenkrantz wrote:
> If it's in CVS, then when you do a commit, you get the generated
> test_char.h. Not what we want...
Yeah, that's another problem.
> I think we could just add a blank copy to the tarball for 2.0.22 (i.e.
> not to CVS). Jeff or someone who cares about supporting system make on
> Solaris 7 can come up with some clever way, or we just always add
> a blank copy of that file to the tarball.
Done, begrudgingly. :-/ We have to come up with a more permanent
solution--as it is, if you check out APACHE_2_0_22 from CVS, you don't get
the file, but if you unzip the tarball, you do [granted, it's a zero-byte
file at that point]. Still, something rubs me the wrong way about the
discrepancy.
--Cliff
--------------------------------------------------------------
Cliff Woolley
cliffwoolley@yahoo.com
Charlottesville, VA