Re: [gentoo-dev] Keywording log

[Eldad Zack]

On Saturday 06 November 2004 21:13, George Shapovalov wrote:
> On Saturday 06 November 2004 07:34, Eldad Zack wrote:
> > > It's called ChangeLog. Everyone except vapier uses it.
> > Somehow I suspected this will be your reply :)
> > I meant something that can be ACURATLY PARSABLE. (hint: Changelog isn't)
> It is, at least supposed to. This very point (the parsability of
> ChangeLogs) was discussed already some time ago and that was the very
> reason we have ChangeLog format as we have it now. We are even supposed to
> use echangelog for still the same purpose :), or be very accurate when we
> hand-edit it (ensure that you maintain formatting!).

No problem with echangelog: the problem is that if I want to find out WHEN was 
an ebuild keyworded ~alpha or stable on sparc, I can't tell.
Except reading the ChangeLog.
But I'd like to be able to have a QA tool to do this for me.
Just look at aliz's script: The "age" of the ebuild is incorrect in some 
cases:
Say I have a package in portage for a year without a change.
Then someone marks it ~amd64 - it catches it as if it is ~amd64 for a year. 
Not reliable at all.

> Thus it does not make sense to create a new file that would be a subset of
> ChangeLog. Realistically it will end up no more parsable than ChangeLogs
> now. In other words wy don't we fix/finish work on what is already accepted
> solution instead of making more decisions and ignoring them later..

I would actually be abit against using yet-another-file to describe it, and it 
would makes more sense to extend metadata for this purpose.
ekeyword can be modified to write the data.

> Oh, you can also use "cvs log". Sometimes (not that rarely in fact) it
> gives you more comments than ChangeLog.

Again, unreliable. Everyone punches variations of what occurs to them in that 
moment. 

-- 
Eldad Zack <eldad@g.o>
Key/Fingerprint at pgp.mit.edu, ID 0x96EA0A93