[m-rev.] diff: enable par grades on MacOS X

[Peter Ross]

On Mon, May 05, 2003 at 09:53:37PM +1000, Fergus Henderson wrote:
> On 05-May-2003, Peter Ross <pro at missioncriticalit.com> wrote:
> > Is it sufficient to issue the following commands to upgrade to the
> > latest version?
> > 
> >         zcat gc6.2alpha4.tar.gz | tar -xvf -
> >         cd gc6.2alpha4
> >         cvs import -I \! mercury/boehm_gc gc gc_6_2alpha4
> 
> No.
> 
> Firstly, that only imports the new version onto the vendor branch
> (well, sort of -- but see below).  To include it in the main branch,
> you then have to merge the changes on the vendor branch back onto the
> main branch.
> 
> Secondly, CVS has a design bug which means that actually the initial
> import will affect the default branch, for *some* files.  So unless you
> take special action (e.g. using "cvs admin -b" to reset the default branch
> for the affected files), the main branch will be in an inconsistent state
> until the vendor branch changes have been merged onto the main branch.
> In other words, doing just the command above will probably stop the
> nightly builds working.
> 
> See the thread at
> http://mail.gnu.org/archive/html/info-cvs/2001-05/msg00088.html for
> details.
> 
Oh dear, I am not sure what has happened now, but this is the order of
the commands that I have executed.

# zcat gc6.2alpha4.tar.gz | tar -xvf -
# cd gc6.2alpha4
# cvs import -I \! mercury/boehm_gc gc gc_6_2alpha4

# cd mercury/boehm_gc

# cvs update .
cvs server: Updating .
P BCC_MAKEFILE
P mallocx.c
P win32_threads.c
cvs server: Updating Mac_files
cvs server: Updating cord
cvs server: Updating cord/private
cvs server: Updating doc
U doc/Makefile.am
U doc/Makefile.in
U doc/gcinterface.html
U doc/leak.html
U doc/scale.html
cvs server: Updating include
U include/Makefile.am
U include/Makefile.in
U include/gc_config_macros.h
P include/gc_cpp.h
cvs server: Updating include/private
cvs server: Updating tests

# cvs admin -b BCC_MAKEFILE mallocx.c win32_threads.c

# cd doc

# cvs admin -b Makefile.am Makefile.in gcinterface.html leak.html scale.html

# cd ../include

# cvs admin -b Makefile.am Makefile.in gc_config_macros.h gc_cpp.h

Which from what I read above meant that everything would be back into a
consistent state.

I then did

# cd gc_6_2alpha4/boehm_gc

# cvs update -j gc:yesterday -j gc .

and proceeded to do the merge.  However I notice that the merge is very
strange.  I think what has happened is that by doing the cvs admin
before cvs update, I have managed to confuse the repository right
royally, which means that the merge for mallocx.c and win32_threads.c
seems to have deleted lines which they shouldn't have and so on.

Does anyone have any suggestions for how to fix this?  I will look into
it now, but if things are broken, this could be a reason why.
--------------------------------------------------------------------------
mercury-reviews mailing list
post:  mercury-reviews at cs.mu.oz.au
administrative address: owner-mercury-reviews at cs.mu.oz.au
unsubscribe: Address: mercury-reviews-request at cs.mu.oz.au Message: unsubscribe
subscribe:   Address: mercury-reviews-request at cs.mu.oz.au Message: subscribe
--------------------------------------------------------------------------