Re: Is there a limint on the number of branches in CVS?

[Todd Denniston]

Jim Hyslop wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Larry Jones wrote:
>
> > Jim Hyslop writes:
<SNIP>
> > > Out of curiosity, has anyone tested what happens if the revision numbers
> > > roll over?
> >
> >
> > I suspect all hell breaks loose.  :-)
>
>
> I suspect you're right.

Would either of you have a better plan for testing the roll over than to do 
something like the following script?
With _an_ unmodified CVS I believe it would take ~18.5 hours to get to 65534, 
to get to 2147483647 it would only take ~68 years by just checking in new 
revs. So we may have a little time left to fix the problem. :)

BTW, looks to me like we still did not have a problem at 4294967294 on a Linux 
system running on a Intel(R) Xeon(TM) CPU 1.50GHz.


#! /bin/bash

BASEWD=/tmp/testcvs
REPODIR=$BASEWD/repo
CVSROOT=:fork:$REPODIR

BIT16=65534
BIT32POS=2147483647
BIT32ALL=4294967294

#setup
mkdir -p $REPODIR/D1/D2
cvs init

PopAVers()
{
  COUNTER=`expr $COUNTER + 1`
  date >> F1.txt
  date >> D2/F2.txt
  cvs commit -m "Commit #$COUNTER" F1.txt D2/F2.txt
}
TwoMore()
{
  echo $COUNTER
  PopAVers
  echo $COUNTER
  PopAVers
  echo $COUNTER
}

COUNTER=1
cd $BASEWD
cvs checkout D1
cd $BASEWD/D1
date > F1.txt
date > D2/F2.txt
cvs add F1.txt D2/F2.txt
cvs commit -m "Commit #$COUNTER" F1.txt D2/F2.txt

for i in $BIT16 $BIT32POS $BIT32ALL
do
  date >> F1.txt
  date >> D2/F2.txt
  COUNTER=$i
  cvs commit -m "Commit #$COUNTER" -r1.$COUNTER  F1.txt D2/F2.txt
  cvs update -A
  TwoMore
  TwoMore
  echo " "
  echo " Still functioning at counter($COUNTER)?"
  echo " "
done

--
Todd Denniston
Crane Division, Naval Surface Warfare Center (NSWC Crane)
Harnessing the Power of Technology for the Warfighter