Eric Blake [EMAIL PROTECTED] writes:
According to Karl Berry on 6/30/2007 12:02 PM:
Since modification of these files is not allowed by the consumer,
It's always been allowed to reformat the licenses. What's not allowed
is to change the words.
So, just like had to be done with
Eric Blake [EMAIL PROTECTED] writes:
--local-dir=local
local/lib/version-etc.c.diff
local/doc/gpl-3.0.texi.diff
or
--local-dir=local
local/version-etc.c.diff
local/gpl-3.0.texi.diff
How about:
local/lib_version-etc.c.diff
local/doc_gpl-3.0.texi.diff
or
local/lib+version-etc.c.diff
I doubt this would ever be triggered, (a user belonging to 2^31 groups
is unlikely, to say the least), but I'm fixing it regardless:
Close the group DB even when failing with 2^31 or more members.
* lib/getugroups.c (getugroups): Don't return without calling endgrent.
Index:
I have installed the following patch:
2007-07-05 Sergey Poznyakoff [EMAIL PROTECTED]
* build-aux/bootstrap (TP_URL,get_translations): Update to use
the new TP address.
(usage): Fix typo
(gnulib_mk): New variable.
Index:
Don't let endgrent clobber errno, no matter how improbable.
* lib/getugroups.c (getugroups): Save and restore errno around
endgrent call.
Index: lib/getugroups.c
===
RCS file:
Hello,
When converting library names to Makefile variables, gnulib-tool
forgets to convert dashes. The following patch fixes it. It also fixes
unneeded duplication of --avoid options, notable in `# Reproduce by:'
headers in generated Makefile.ams.
No objections?
Regards,
Sergey
2007-07-05
Eric Blake wrote:
For an example, with
just gpl-3.0.texi.diff and version-etc.c.diff, the difference is between:
--local-dir=local
local/lib/version-etc.c.diff
local/doc/gpl-3.0.texi.diff
or
--local-dir=local
local/version-etc.c.diff
local/gpl-3.0.texi.diff
The latter introduces
Jim, Bruno, thank you. Gathering the bits together, I have rewritten the
thing as follows:
Index: gnulib-tool
===
RCS file: /cvsroot/gnulib/gnulib/gnulib-tool,v
retrieving revision 1.245
diff -p -u -r1.245 gnulib-tool
---
Sergey Poznyakoff [EMAIL PROTECTED] wrote:
Jim, Bruno, thank you. Gathering the bits together, I have rewritten the
thing as follows:
Index: gnulib-tool
...
- libname_upper=`echo $libname | LC_ALL=C tr 'a-z' 'A-Z'`
+ libname_upper=`echo $libname | LC_ALL=C tr 'a-z-' 'A-Z_'`
Looks
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
According to Bruno Haible on 7/5/2007 4:20 AM:
This is a bit better, but maybe '%' or ':' are better separators.
'%', maybe. ':', absolutely not. Using a colon will foul up CVS
checkouts to picky file systems (such as anything on Windows).
Jim Meyering [EMAIL PROTECTED] ha escrit:
I know you didn't change the single quotes (maybe it's Bruno's
stylistic preference to use them?), but they are superfluous.
There's less syntax this way:
libname_upper=`echo $libname | LC_ALL=C tr a-z- A-Z_`
Yeah, but the quoted variant
Jim Meyering wrote:
I know you didn't change the single quotes (maybe it's Bruno's
stylistic preference to use them?)
Yes, it's my stylistic preference.
but they are superfluous. There's less syntax this way:
libname_upper=`echo $libname | LC_ALL=C tr a-z- A-Z_`
But it's not more
Bruno Haible [EMAIL PROTECTED] wrote:
Jim Meyering wrote:
I know you didn't change the single quotes (maybe it's Bruno's
stylistic preference to use them?)
Yes, it's my stylistic preference.
but they are superfluous. There's less syntax this way:
libname_upper=`echo $libname |
Eric Blake [EMAIL PROTECTED] wrote:
When configuring with -Werror, the gettimeofday module decides that
cygwin's gettimeofday is non-compliant because a warning is made fatal:
Hi Eric,
Isn't there general consensus that using -Werror at configure-time
is best avoided, for precisely this
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
When configuring with -Werror, the gettimeofday module decides that
cygwin's gettimeofday is non-compliant because a warning is made fatal:
configure:13103: checking for gettimeofday with POSIX signature
configure:13135: gcc -std=gnu99 -c -gdwarf-2
Benno Schulenberg [EMAIL PROTECTED] wrote:
Jim Meyering wrote:
You can argue that the translation project should not publish
such a .po file, and I agree that upstream (in the TP
.po-checking code) is the right place to fix this, but can you
get them to fix it quickly
I've run the following
Jim Meyering wrote:
You can argue that the translation project should not publish
such a .po file, and I agree that upstream (in the TP
.po-checking code) is the right place to fix this, but can you
get them to fix it quickly
I've run the following on all files in the latest/ directory that
Net::CDPhttp://search.cpan.org/src/MCHAPMAN/Net-CDP-0.09/libcdp/
OPeNDAP libdap http://scm.opendap.org:8090/svn/trunk/
I wrote these two. The Net::CDP author already replied saying he is
willing to relicense libcdp under GPLv3. No word from the OPeNDAP
people.
gpg
OPeNDAP is under the LGPL, not GPL.
Right, but the same issue arises.
And if I remember well it is under LGPL 2 or above.
I could not find any obvious statement. The source files I looked at
didn't have any copyright/license info at all. Sigh.
On Thu, Jul 05, 2007 at 05:04:43PM -0500, Karl Berry wrote:
OPeNDAP is under the LGPL, not GPL.
Right, but the same issue arises.
And if I remember well it is under LGPL 2 or above.
I could not find any obvious statement. The source files I looked at
didn't have any
Eric Blake wrote:
cygwin's sys/time.h declares struct
timeval, then includes sys/select.h, which in turn includes sys/time.h
to guarantee struct timeval is declared no matter which header was
included first. But once gnulib has decided that cygwin needs the
replacement, it turns into the
Karl Berry wrote:
The Net::CDP author already replied saying he is
willing to relicense libcdp under GPLv3. ...
Werner has already updated the gnupg source repository to GPLv3.
Good! To summarize: All GPLed (non-LGPLed) packages that use gnulib are
fine with GPLv3.
Bruno
Btw, I don't follow what would give me the right to reformat the
license though, the copying information seems pretty clear to me:
I don't have a policy document I can point to, but I know that is the
accepted practice. Our web pages are also usually verbatim only, but
verbatim doesn't
is necessary to s/unnumbered/appendixsec/ if including the GPLv3 in an
Yeah, the discrepancy between the FDL and the GPL seems like a bug.
I'll ask about that change.
or if texinfo could support some other subsectioning keyword that
works regardless of whether the parent is a main
Patrice Dumas wrote:
It is specified in all the source files! Did you had a look at
http://www.opendap.org/pub/source/libdap-3.7.8.tar.gz
Glad to see this see. I had thought it was LGPLv2.1 because the README says
COPYRIGHT INFORMATION
The OPeNDAP DAP library is copyrighted using the
You mean we extend the GPL/LGPL trick, where the actual copyright
of the files is different from the copyright notice in the files?
I've never understood that trick (even though you've explained it to
me before, sorry). On the face of it, it does not seem legal.
Brett asked to be
Paul Eggert wrote:
We can implement a --gplv3 parameter om gnulib if you
don't want to have GPLv2 mentioned in your sources.
That sounds like a good idea, thanks. The default, though, should be
GPLv3, and we can implement a --gplv2 for the old-fashioned projects.
Any objections to this
Jim Meyering wrote:
I have no problem here. But my be.po file has
already been merged, so it's no longer the same as yours.
Yes. msgmerge can, in such cases, add 'fuzzy' markers, to avoid failures
of msgfmt -c.
Bruno
Karl Berry wrote:
I've never understood that trick (even though you've explained it to
me before, sorry).
It is documented in the README, in more detail in doc/gnulib-intro.texi, and
you find the motivation in
http://lists.gnu.org/archive/html/bug-gnulib/2004-09/msg00131.html
On the face of
getugroups could fail (due to getgrent), but the caller
would have no way to know that.
* lib/getugroups.c (getugroups): Detect getgrent failure.
Adjust comment to reflect reality: this function may return -1.
Index: lib/getugroups.c
30 matches
Mail list logo