Dailydave mailing list archives
Re: Another anonymized posting.
From: Rodney Thayer <rodney () canola-jones com>
Date: Wed, 09 Jun 2004 16:15:26 -0700
At 04:02 PM 6/9/2004 -0700, Matt Hargett wrote:
Rodney Thayer wrote:At 06:26 PM 6/9/2004 -0400, Dave Aitel wrote:anonymized posting please: BTW, the way things are going, I'd consider switching to svn.and why? what was the redeeming social value in CVS to begin with? automerge, which never worked? claims of tree corruption, which were never substantiated? is there some reason you have an intense urge to have a server-based source tree mechanism to slow down development? what's wrong with RCS?Auto-merge
Automerge is the most incredibly rock-stupid software engineering technique since programmers used line printers to generate pin-ups. Letting a machine decide that you had two team members pissing on the same source module is good - for causing exploitable code.
works most of the time with small non-cowboy development teams who have componentized things decently and therefore don't do double edits.
Professional teams with poor enough team management that they have double edits deserve to be reamed by the likes of Dr. Aitel and Brother Moore.
Clearcase
Clearcase is a baroque piece of shit that is great for 9000 person development teams run by fist-pounding clueless engineering executives who'd rather pay 3 cents an hour for shitty offshore code than pay a competent local developer to build things decently.
CVS and RCS are both differential based, which some people really don't like.
They'd rather the client did the diffing, since that's where the smarts for different diffing approaches based on filetype should be done.
What the f. are you talking about? CVS and RCS both do the diff locally, as far as I can tell. And if you don't based source control on diffs you never notice the changes ... see previous rantage on how you get exploitable code.
Of course, if you don't need diffs, branching, labels, post-commit triggers, etc then just tar up your source every x hours and untar in case of disaster.
Post-commit triggers are a sign of poor tool choices, don't go there. labels and diffs are what you need. branching is cute but avoidable in real software development teams. that's what well defined component interfaces are for. branching is typically used by engineering teams to stonewall/avoid fixing outstanding bugs (read: exploitable code.) The reason I bothered to speak up is that in my opinion, use of these ornate overengineered source code control systems is not a replacement for developing software that works and has minimal flaws. However, modern software develoment teams will never turn down a chance to go with a new tool if it allows them to write sloppy code. Oh, well, more work for Dave and the snake. _______________________________________________ Dailydave mailing list Dailydave () lists immunitysec com http://www.immunitysec.com/mailman/listinfo/dailydave
Current thread:
- Another anonymized posting. Dave Aitel (Jun 09)
- Re: Another anonymized posting. Rodney Thayer (Jun 09)
- Re: Another anonymized posting. Matt Hargett (Jun 09)
- Re: Another anonymized posting. Rodney Thayer (Jun 09)
- Re: Another anonymized posting. Matt Hargett (Jun 09)
- Re: Another anonymized posting. H D Moore (Jun 09)
- Re: Another anonymized posting. ned (Jun 09)
- <Possible follow-ups>
- Re: Another anonymized posting. ninjatools (Jun 09)
- Re: Another anonymized posting. Rodney Thayer (Jun 09)
