Messages in this thread | ![/](/images/icornerl.gif) | | Subject | Re: Is the BitKeeper network protocol documented? | From | Valdis.Kletnieks@vt ... | Date | Mon, 20 Jan 2003 15:32:53 -0500 |
| |
On Mon, 20 Jan 2003 11:43:48 PST, David Schwartz said:
> Checking source code out of a repository is an obfuscatory act that > separates the raw source code from the rationale for that source > code. It's equivalent to stripping comments. The GPL does not allow
So is shipping the source without the transcript of the kernel developer's conference, because then you're stripping out some of the design rationale.
So is shipping the source without a neuron dump of the programmer - let's face it, we've ALL looked at code and said "What WERE they thinking?", and therefor a neuron dump would be part of the *preferred* format.
You seem determined to obfuscate the issue by confusing the *SOURCE* that actually gets modified, and metainformation used to keep TRACK of the source.
Quick sanity test:
Do people actually modify the BK repository, or do they check it out so they can actually modify *THE CHECKED OUT TREE*? Last I heard, people actually did edits on the source tree, and they used gcc (or whatever compiler) on the source tree. Then they make diffs between the old tree and new tree and send them to Linus.
Seems to *ME* that since almost all of the source code was actually created in a '(vi|emacs) / make / gcc / diff' environment, that is the PREFERRED format for making modifications.
Don't confuse the source tree with metainformation, or you'll end up having to carry around inode information. Lest you think I'm joking, consider the fact that the original Crowther&Woods Adventure game was called 'ADVENT.FOR', and the case and number of chars was actually significant information....
[unhandled content-type:application/pgp-signature] | ![\](/images/icornerr.gif) |