Re: Keyword substitution command options override binary status forrepository files

[Susanne Bertling]

Hi Dieter,

> The "keyword2-20k" test has unexpected behaviour. It outputs
> two "U binfile.dat" lines. This may be a bug in updating
> binary files exposed but not caused by my patch. I leave the
> analysis to someone more familiar with "cvs" than I am.

There are the following commands in sanity.sh (keyword2-16 - keyword2-20):
cvs update -A (binfile.date has revision 1.1)
cvs tag -b branch2
cvs update -r branch2
changes to binfile.dat
cvs ci (new revision is 1.1.4.1)
cvs update -A -kk -j branch2
At the last command cvs first checks out the head revision (1.1) of 
binfile.date, the first 
'U binfile.dat' then it has to merge the differences between the greatest 
common ancestor 
of the head revision and the last revision into the head revision. But the 
greatest common 
ancestor is revision 1.1, so revision 1.1.4.1 is the result of the merge.
The second output 'U binfile.dat' indicates, that there was a second checkout 
of the file 
(this time of revision 1.1.4.1).
As an explanation in 'update.c' you can read the comment
/* Traditionally, the text file case prints a whole bunch of
scary looking and verbose output which fails to tell the user
what is really going on (it gives them rev1 and rev2 but doesn't
indicate in any way that rev1 == vn_user). I think just a
simple "U foo" is good here; it seems analogous to the case in
which the file was added on the branch in terms of what to
print. */

I think the output isn't very informative about what is going on, but the test 
case ist handled 
correctly. When you use the nonmodified cvs you get the same output if you 
replace the option 
-kk by -kb.

Susanne
P.S.: I'd like the patch to be included in cvs, too.
______________________________________________________________________________
Wo gibt es den besten Spam-Schutz? Laut ComputerBild 15-03 bei
WEB.DE FreeMail - Deutschlands beste E-Mail - http://s.web.de/?mc=021122