ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stefan Bodewig <bode...@bost.de>
Subject Re: icvs integration
Date Tue, 04 Jul 2000 08:59:00 GMT
>>>>> "MD" == M Dietrich <mdt@diwa-gmbh.de> writes:

 MD> i wrote this before but it probably get lost.

No it didn't, I responded to it. My original response is below.

Stefan

------------------

>>>>> "MD" == M Dietrich <mdt@diwa-gmbh.de> writes:

 MD> Hello, i found the following comment in Cvs.java:

 MD> // XXX: we should use JCVS (www.ice.com/JCVS) [...]

This comment has lead Erik Meade <emeade@geekfarm.org> to implement a
JCVS task and post it here a while back (beginning of June). 

Maybe we should come to a decision and remove the comment after that
(or change the Cvs task depending on the outcome).

A JCVS task could only be an optional task as we can't bundle JCVS
with Ant (the GPL problem) - and user who don't want JCVS couldn't
compile Ant if we made it a core part of it.

Apart from that my main objection against replacing the current Cvs
task with a JCVS implementation was, that JCVS didn't support the
:local: and :ext: protocols. Erik has told the list that the addition
of those is planned so sometime in the future JCVS could be a full
replacement of the command line version, but it is not right now.

My proposal: Encourage an optional JCVS based task and remove the comment
from Cvs.java.

As soon as JCVS can do everything the command line version can (and I
*need* local as well as ext) we can make the Cvs task optional as
well. The bootstrap process would compile in the JCVS version if it
found JCVS and use the command line based version otherwise.

Comments?

Stefan


Mime
View raw message