You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@subversion.apache.org by Jim Whitehead <ej...@cse.ucsc.edu> on 2001/07/13 00:58:36 UTC

Suggestion for cvs->svn converter

First, congratulations on getting the file system deltified! I'm eagerly
awaiting the day when I can switch over to using subversion, and I can tell
that day is not too far off.

I worked at Continuus for a summer and did some development on their
Migration tool, which sucked code into the Continuus repository from SCCS or
RCS repository files. One thing that was very important in the tool was to
provide a status indicator that gave an estimate of the time remaining in
the conversion activity. Since it took appx. 1 second per revision (in
1994), importing a large project could take hours. It was handy for people
to know that they could fire off the import, than come back later in N
hours. Machines are much faster now, but still, I imagine the cvs->svn
converter might still take O(minutes) to finish.

Logging was also important, since it allowed the admin. in charge of the
import to know which revisions experienced problems during the import
operation.

I suppose it would also be handy if the conversion utility used a plug-in
architecture, so people could develop conversion plug-ins for multiple
version control systems, rather than just limit the utility to CVS.

My $0.02,

- Jim


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org

Re: Suggestion for cvs->svn converter

Posted by Karl Fogel <kf...@collab.net>.
"Jim Whitehead" <ej...@cse.ucsc.edu> writes:
> I worked at Continuus for a summer and did some development on their
> Migration tool, which sucked code into the Continuus repository from SCCS or
> RCS repository files. One thing that was very important in the tool was to
> provide a status indicator that gave an estimate of the time remaining in
> the conversion activity. Since it took appx. 1 second per revision (in
> 1994), importing a large project could take hours. It was handy for people
> to know that they could fire off the import, than come back later in N
> hours. Machines are much faster now, but still, I imagine the cvs->svn
> converter might still take O(minutes) to finish.
> 
> Logging was also important, since it allowed the admin. in charge of the
> import to know which revisions experienced problems during the import
> operation.

+1 on both suggestions.

Really huge, multi-gigabyte repositories, like the RedHat (Cygnus) GNU
tools codebase, are going to take noticeable time to convert -- we are
at least talking about minutes, if not hours.  A progress indicator
and log are necessities.

> I suppose it would also be handy if the conversion utility used a plug-in
> architecture, so people could develop conversion plug-ins for multiple
> version control systems, rather than just limit the utility to CVS.

Sure, just depends how much extra work it is.

-K

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org