>--- Forwarded mail from [EMAIL PROTECTED]
>[EMAIL PROTECTED] on 2000.02.18 16:41:24
>>I'm personnaly stoned by the number of people who turned to cvs due to
>>the efforts of projects like MacCvsPro, WinCvs, jcvs and others (I
>>noticed that according to the mail box of WinCvs.org and the stats on
>>the pages). Just make some querys on the web, you'll see. It is just
>>fair to begin to think a little bit like the regular Windows-Web users
>>who turns to cvs : they want some high level stuff.
>I think both goals are achievable. CVS's client/server protocol is very well
>documented. Those wishing higher level tools can build a layer on top of CVS
>(just look at all the GUI front-ends to CVS).
>For example, those wishing hard locks can wrap CVS such that one command can
>checkout (in the CVS sense) or update and lock (ie "cvs edit -c") in one shot.
This particular scenario is fine for wrappers, because static state can be
gathered while the repository is locked (once, during checkout or update)
and then additional action taken based on that static state. However, there
are actions do not lend themselves to that model (e.g. anything that
specifically operates on the HEAD revision) where wrappers around the
CVS executable will fail without some kind of long transaction mechanism
(i.e., the ability to invoke multiple CVS commands on a static repository,
with the guarantee that no one will modify one of the RCS files that is
of interest to the wrapper).
>--- End of forwarded message from [EMAIL PROTECTED]
