Re: [Cvs-dev] cvs features for gnu savannah

[Mark D Baushke]

Thorsten Glaser <address@hidden> wrote:

> On Wed, 30 Nov 2016, Conrad T. Pino wrote:
> 
> > I volunteer to archive CVS top-of-tree into separate branch and then
> > revert
> > specific ChangeLog commits to bring CVS top-of-tree into a state
> > ready to
> > accept new commits for a future 1.12.14 release.
> 
> Please *DON-1òùT* rewrite history like this. We can find other
> solutions-1ó¦ such as developing 1.12.14 and up on a branch.

I agree with this. Please do NOT rewrite history. We can always commit
forward the reverse diffs to unwind the history if we really need to do
it that way. However, we probably will want to tag the top-of-tree
main.trunk as a memorial-derek-price-tag. It was as far as he got with
trying to have signed commits for the FSF folks that commissioned the
work.

> One more thing-1òêòôòêsomeone suggested merging some unnamed fixes
> from 1.11 into 1.12; if you can point them out to me for forward-
> porting, that would be welcome (I don-1òùt know whatòùs in 1.11 and
> still needed in 1.12).

I do not believe there are any patches actually in the cvs source base
that require this.

$ cvs diff -u -rcvs1-11-x-branch-last-merge -rcvs1-11-x-branch
Index: ChangeLog
===================================================================
RCS file: /cvsroot/cvs/ccvs/ChangeLog,v
retrieving revision 1.692.2.268
retrieving revision 1.692.2.269
diff -u -p -r1.692.2.268 -r1.692.2.269
--- ChangeLog   12 Sep 2008 15:45:21 -0000      1.692.2.268
+++ ChangeLog   19 Jul 2009 04:03:34 -0000      1.692.2.269
@@ -1,3 +1,8 @@
+2009-07-18  Derek R. Price  <address@hidden>
+
+       * NEWS: Note timestamp fix for systems with imperfectly
synchronized
+       time() & gettimeofday() functions.
+
 2008-09-12  Derek Price  <address@hidden>

        * NEWS: Note fix for bug #17565.
$ 

So, the 'missing' patches are those that may have been submitted to the
bug-cvs list or those that have been provided against cvs 1.11.23 in
other distributions.

There were also a bunch of NetBSD coverity bug fixes out there which
could/should be fixed.

Conrad T. Pino write:

> 1) At this time everyone seems to agree CVS top-of-tree (MAIN) is
>    unacceptable having incomplete changes no one wants to complete; GPLV3+
>    in particular.

The current state of the top-of-tree main.trunk does NOT have any GPLv3+
code in it as yet. Work stopped before the GNULIB update to GPLv3+
had as yet commenced.

I suggest that there be a normal tag taken to mark the state of cvs
1.12.13.1 which was the main.trunk with some manner of tagname. This
would allow someone to later create a branch at that point and work
forward on Derek's design if they so desired.

That said, there is much in the 1.12.13.1 which I found very useful and
we may wish to revisit in the future.

  * The sanity.sh script was much improved over what was used by
    1.11.23.

  * The commit triggers flexibility was better in general. Sadly, the
    commitid code while very desirable was never made into a first-class
    database tracked token.

  * Some of the CVSNT features were shared between the two projects.
    (e.g., :extssh:)

  * The FreeBSD 'blame' synonym for 'annotate'

  * Some of the mmap enhancements.


SOme of the things not done, but would be nice if they could be done
include upgrading the 'diff' implementation baseline (from 2.7 to
2.8.1).

        -- Mark