Re: [PATCH] t9400: Add some tests for checkout

From: Frank Lichtenheld <frank@lichtenheld.de>
Date: Oct 4, 2007, 19:12 UTC
Message-ID: <20071004191217.GC31659@planck.djpig.de>
In-Reply-To: <Pine.LNX.4.64.0710041921490.4174@racer.site>

On Thu, Oct 04, 2007 at 07:24:05PM +0100, Johannes Schindelin wrote:

Show 12 quoted lines

> On Thu, 4 Oct 2007, Frank Lichtenheld wrote:
> 
> > +# clean up
> > +rm -fr cvswork2
> > +rm -fr "$SERVERDIR"
> > +cd "$WORKDIR" &&
> > +git clone -q --local --bare "$WORKDIR/.git" "$SERVERDIR" >/dev/null 2>&1 &&
> > +GIT_DIR="$SERVERDIR" git config --bool gitcvs.enabled true &&
> > +GIT_DIR="$SERVERDIR" git config gitcvs.logfile "$SERVERDIR/gitcvs.log" ||
> > +exit 1
> 
> Should this not be in a test_expect_success, too?

Since I do this several times and since it is easier to see what tests it belongs to if it isn't buried in one of them, I would say "no".

Show 16 quoted lines

> > -#------------
> > -# CVS UPDATE
> > -#------------
> > +#--------------
> > +# CVS CHECKOUT
> > +#--------------
> > +test_expect_success 'cvs checkout failure (HEAD)' \
> > +  'if GIT_CONFIG="$git_config" cvs -Q co -d cvswork2 HEAD >cvs.log 2>&1
> > +   then
> > +     echo unexpected cvs success
> > +     false
> > +   else
> > +     true
> > +   fi &&
> 
> How about "! GIT_CONFIG..." instead of the "if..then..else..fi" ?

I don't really care. IIRC I took the idiom from another testfile. Several other tests in this file already use it, too. So unless Junio prefers I change all occourences, I will not.

Show 5 quoted lines

> > +   cat cvs.log | grep -q "not a branch" &&
> > +   test ! -d cvswork2'
> > +rm -fr cvswork2
> 
> Again, for consistency, I'd include this in the test case.

Again, I don't really care. But for consistency with the rest of the file I will only change it if I'm asked to change all occourences.

Gruesse,

-- 
Frank Lichtenheld <frank@lichtenheld.de>
www: http://www.djpig.de/