Recommended approach for converting from CVSNT to Mercurial?

greg fenton greg.fenton at gmail.com
Wed Jan 13 22:11:23 CST 2010


Just to be clear, that is the total of my entire CVS repository which
comprises many modules.  I believe that cvs2hg should be run on a
per-module basis, right?

In that case the numbers are much broken down.  And to be honest, I'm
sure I could convince the development team to drop 70% of the modules
or simply leave them in CVS since they are almost completely unused
these days.

I am running cvs2hg overnight in a loop over my list of modules (bash
script below).  I'll let you know what my runtimes are (I don't have
time to set up other monitoring tonight...I'll do so in a future
re-run).

Thanks for everyone's input to date!
greg.fenton
--
greg.fenton ^_at_^ gmail (yes, dot com)

#!/bin/bash
#----------------------------------------------------------
cvs_repo=/home/greg/cvs_repository/code
hg_repo=/home/greg/hg_repository
CVS2HG=~/nexj/cvs2svn/cvs2hg
LOG=~/cvs2hg_convert.out

cd ${cvs_repo}

for i in *
do
    echo ""
    echo ""
    echo ""
    echo "=====  [`date`] ${i}"
    echo ""
    echo ""
    $CVS2HG --encoding=cp1252 --hgrepos $hg_repo/${i} ${i}
    echo ""
    echo ""
    echo "=====  [`date`] ${i}  -- DONE"
done | tee -a ${LOG}
#----------------------------------------------------------





On Wed, Jan 13, 2010 at 22:14, Greg Ward <greg-hg at gerg.ca> wrote:
> On Wed, Jan 13, 2010 at 5:00 PM, greg fenton <greg.fenton at gmail.com> wrote:
>> I have a mature CVSNT repository that I'd like to convert to Mercurial.  It
>> is approximately 16G, 100000 files, with two archives being ~1G in size.
>
> Wow.  That is big -- about 4x more files and 16x more bytes than I am
> currently feeding into cvs2hg.  My conversions (26,000 files, < 1 GB,
> 105,000 changesets) are taking ~20 hours.  Hope you have fast hardware
> and lots of memory!
>


More information about the Mercurial mailing list