30 messages in com.perforce.perforce-userYour 97.3 Input Solicited
FromSent OnAttachments
Robe...@perforce.com18 Jun 1997 15:01 
Carr...@perforce.com18 Jun 1997 15:18 
Rick...@charger.engr.sgi.com18 Jun 1997 16:02 
Nick...@navio.com18 Jun 1997 16:08 
Gerd...@BITart.com18 Jun 1997 16:13 
Jeff...@uplanet.com18 Jun 1997 16:59 
Greg...@relay.engr.SGI.COM18 Jun 1997 17:00 
BobA...@weblogic.com18 Jun 1997 17:16 
Matt...@geoworks.com18 Jun 1997 18:40 
Jeff...@uplanet.com18 Jun 1997 19:30 
Paul...@radstone.co.uk19 Jun 1997 01:09 
Robe...@perforce.com19 Jun 1997 09:27 
Jeff...@uplanet.com19 Jun 1997 09:33 
Johnston19 Jun 1997 09:37 
EdHe...@webtv.net19 Jun 1997 11:33 
Bill...@firepower.com19 Jun 1997 11:51 
Tony...@illustra.com19 Jun 1997 12:18 
Anth...@jpmorgan.com19 Jun 1997 12:21 
Matt...@geoworks.com19 Jun 1997 15:50 
Nave...@vinayak.engr.sgi.com19 Jun 1997 16:33 
Davi...@isltd.insignia.com20 Jun 1997 01:20 
Davi...@isltd.insignia.com20 Jun 1997 01:52 
# Cowham20 Jun 1997 02:20 
Anth...@jpmorgan.com20 Jun 1997 06:30 
Matt...@geoworks.com20 Jun 1997 10:19 
Matt...@geoworks.com20 Jun 1997 10:24 
Tony...@illustra.com23 Jun 1997 13:26 
Sush...@datatools.com23 Jun 1997 14:12 
Jeff...@mercury.xionics.com24 Jun 1997 09:00 
Ames...@cloudscape.com24 Jun 1997 10:10 
Subject:Your 97.3 Input Solicited
From:Anth...@jpmorgan.com (Anth@jpmorgan.com)
Date:06/20/1997 06:30:16 AM
List:com.perforce.perforce-user

David Tock wrote:

For P4CLIENT to have a default is I think a bit unhelpful ...

Can't agree with this one. We do cross platform development [...] so having the client default to the machine name is ideal. I guess I could go for P$CLIENT=$USER-$HOSTNAME, but just $USER would really cause me (and many of my colleagues) grief.

OK. All I was trying to say was that the hostname default only fits one environment, and only made sense to me once I realized that it was designed for setups like yours. At that point, everything was clear.

Lastly, everything with Perforce felt a bit irrevocable.

It is. Good. [...] No thanks. Full audit trail please - warts and all...

Hmm. I do understand your perspective. But each to their own here, I think. I would have felt more comfortable if, for example, I could create a new repository, copy only valuable files (and their history) across, and then completely destroy the old repository. Without this, it feels like either you get the structure right first time, or else you've got to learn a heck of a lot about branching and deleting in order to undo the damage.

An audit trail is fine once the system is going full steam and there's a bit of expertise in using it, but at the start one needs to experiment. And I just don't have time to experiment with toy code - the system goes live on real code from day 1. Yes, I admit it's playing with fire. But that's why I'd want a low-level fire-extinguisher available.

One other thing. I was surprised to find that I couldn't map my RCS tree through to Perforce. I know there's a rcs2perf script (or somesuch name) but, if you have a look at what it does, it's a bit limited. I would have liked to be able to add RCS trees into a depot, and that didn't seem possible.

Regards

Anthony