cvs-dev
[Top][All Lists]
Advanced

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

[Cvs-dev] Re: [Cvs-cvs] ccvs ChangeLog NEWS src/ChangeLog src/import.c .


From: Derek Price
Subject: [Cvs-dev] Re: [Cvs-cvs] ccvs ChangeLog NEWS src/ChangeLog src/import.c ...
Date: Mon, 18 Jun 2007 10:06:05 -0400
User-agent: Thunderbird 1.5.0.12 (Windows/20070509)

Mark D. Baushke wrote:
Hi Derek,

I have problems on GNU/Linux running with your latest patches.
To reproduce, try the following:

  sh sanity.sh -r `pwd`/cvs import-CVS

There will be an error message in check.log looking like this:

cvs [import aborted]: reading CVS/Tag: Not a directory
exit status was 1
FAIL: import-CVS-2

Sorry, I was in a hurry and skipped running remote checks. I'll look into it.

In other news...

I have a tree wherein I have an update to use the latest GNULIB files
(patch attached). There are a few problems with it regarding the
windows-NT port...

For the windows bulid of CVS, there are problems with the GNULIB use
of this idiom:

#if @HAVE_INCLUDE_NEXT
# include <foo.h>
#else
# include @ABSOLUTE_FOO_H@
#endif

The configure script subs the @HAVE_INCLUDE_NEXT@ with a 1/0, but I forget the rationale for preferring this over a macro set in config.h. It may have had something to do with GNULIB headers replacing system headers needing to be able to be sourced before/without including config.h.

Anyhow, you can look in windows-NT/Makefile.am. Conrad Pino and I came up with some targets to build stdint.h & stdbool.h from the GNULIB versions. The procedure should be similar for these other files. Of course, with this many files to generate, it might be worth adapting the windows-NT/mkconfig.pl script to generate them in a manner similar to how windows-NT/config.h is being generated since there are bound to be duplicated keys.

I don't believe that I understand the
pathname for the various ABSOLUTE_FOO_H values to be used... assuming
that there is a windows standard install path for them.

This I'm not sure about. I would have guessed that it would be tricky to get this right under Windows since the headers might be installed at any number of locations, but it looks like Conrad just chose to replace ABSOLUTE_STDINT_H with the empty string. Again, I am unsure of the rationale, but if it works, it may work for other files.

Any suggestions? In the worst case, I figure I could just commit the
patch and let someone else that has a window-NT box help sort things out
later to come up with a suitable replacment for each of the .h files as
needed.

If you like. I believe the windows-NT port is currently disabled anyhow. No one has been supporting it lately.

Regards,

Derek
--
Derek R. Price
Solutions Architect
Get CVS support at Ximbiot <http://ximbiot.com>!
v: +1 248.835.1260
f: +1 248.835.1263
<mailto:address@hidden>




reply via email to

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