bug-cvs
[Top][All Lists]
Advanced

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

Re: CVS/Template files ... requesting opinion/clue


From: Derek Robert Price
Subject: Re: CVS/Template files ... requesting opinion/clue
Date: Fri, 07 Mar 2003 15:38:48 -0500
User-agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.0.2) Gecko/20030208 Netscape/7.02

Mark D. Baushke wrote:

I agree that -s should imply -r, so my patch will do that. I will also
have -c imply -r as why would you need a client to be different than
just putting the path to the cvs under test as the argument.

It should be understood that the -s and -c switches will only make sense
with individual cases or ranges of cases as there have been some changes
that are bug fixes that impact the output of the various commands.


Hrm. Why allow a -c at all? Since the script always requires a path to a CVS executable, why can't that double as the client? That is the way your indirection works out at the moment and when -c and -s are specified, the command argument isn't used..

I think that replacing $testcvs and ${testcvs} with ${clientcvs} should
be considered as a separate patch. I am not sure it is the right thing
to do as using a ${clientcvs} implies client/server testing to me...


Well, kinda, but since it seems to be a redundant specification to me, it makes sense to call whatever the user execs a client and anything the client execs the server. Coincidentally, the client performs server operations in local mode.

I will document the $clientcvs, $cmdargcvs and $servercvs variables I
have introduced as a part of this Template patch in the TESTS file.


I just documented $remote and $keep in TESTS, though not the naming standard yet.

The lack of consistent $keep exit points is because $keep is a fairly
recent addition to the suite and nobody took the time to revamp the
whole script.  The rule I've been using is that I add $keep anytime I
work on a test if it isn't there already, to any new test, and any
time I need it and it isn't there.

Would it make a bit more sense to do something like a shell function?

dokeep() { if ${keep}; then
     echo "Keeping ${TESTDIR} for test case \`${what}' and exiting due to 
--keep"
     exit 0
   fi
}

and then pass in ${TESTDIR} ? So, replace

          if $keep; then
            echo Keeping ${TESTDIR} and exiting due to --keep
            exit 0
          fi

with

          dokeep

which is a bit easier to add to all of the tests. Of course, the option
is really -k rather than --keep so perhaps the comment should change?


If you like. This should be pretty easy to wrap tests in once I start porting the suite into the autoconf autotest tools (Issue #64 <http://ccvs.cvshome.org/issues/show_bug.cgi?id=64>), but who knows how long it'll be before I get to that.

If you're feeling motivated enough to put $keep exit points in all the
tests, by all means feel free.  :)

I might do it if I have time, but I will do it as a separate cleanup
commit.

Incidentally, I'll update sanity.sh to accept a --keep argument in a few moments.

:)

Derek

--
               *8^)

Email: address@hidden

Get CVS support at <http://ximbiot.com>!
--
All those who believe in psychokinesis raise my hand.







reply via email to

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