Subject: the donut rule (was Re: 0.7.8 and ANISfied string functions)
From: Paul Rohr (paul@abisource.com)
Date: Sat Jan 29 2000 - 11:41:47 CST
At 01:37 PM 1/29/00 -0000, James Montgomerie wrote:
>I've basically finished the ANSIfied sting functions now, but a [fairly]
>large patch is required to make the whole tree use them (i.e. to replace all
>the stricmp, strcasecmp etc. with UT_stricmp etc.). Should I hold back for
>0.7.8 to be out before I submit the patch?
Thanks for checking.
If you feel that your patch meets all of the following criteria, feel free
to submit it now:
1. It follows the coding standards.
2. It will apply cleanly against a current view of the CVS trees.
3. You've thoroughly tested it.
4. In your opinion, it's unlikely to destabilize the tree.
5. You're willing to resubmit it if we disagree about #1-4. ;-)
Once it arrives, someone with CVS access will need to review the patch and
decide whether to sneak it in now or play it safe and wait.
Our general rule is that anyone who commits code that leaves the build
broken (without prior notice) on *any* of our supported platforms is
responsible for either:
- getting it fixed promptly and/or
- buying donuts for everyone who lost time because of it
Given cool tools like Bonsai and cvsblame, it's pretty easy to tell who this
"donut rule" applies to. When I've worked on colocated development teams,
this rule (and its associated sweeteners) worked really, really well.
However, this team couldn't possibly be more geographically dispersed, so
this is a more abstract concept than it would be for a colocated team. (Do
virtual donuts taste better than virtul beers?)
Given how close we are to the next release, any big patch like this would
need to look *really* good before I felt comfortable that it wouldn't break
the build or destabilize any of our supported platforms.
My personal goal from now until 0.7.8 is to make extra-sure that nothing
"donut-worthy" happens. ;-)
Paul
This archive was generated by hypermail 2b25 : Sat Jan 29 2000 - 11:36:25 CST