Re: [Gnu-arch-users] archzoom wishlist item

[Harald Meland]

[Mikhael Goikhman]

> On 22 Aug 2004 11:59:32 +0200, Johannes Berg wrote:
>> 
>> On Sa, 2004-08-21 at 19:15 -0400, Miles Bader wrote:
>> > Still, I think it would obviously confuse people if the order they see
>> > everyday in their names was not the same as that presented in a browser.
>> 
>> Sure. But for those who actually know what they're doing, tracking
>> things that essentially have a single version number (like my example)
>> gets much much easier. I'm still not advocating changing the default.
>
> Ok, I implemented this, although I think that transposing branch and
> version is a bad idea. Take a look at:
>
>   http://archzoom.sourcecontrol.net/demo/address@hidden

Although I'm not to sure about this particular way of implementing the
feature, I still think that e.g.

  http://archzoom.sourcecontrol.net/demo/address@hidden

is, for some of the likely uses of an Arch browser, much preferrable
to the non-"?transpose" variant.  For other uses, the non-"?transpose"
variant is preferrable.

(Note that in this "?expand" view, it's actually the sort order which
matters; do you sort on "branch" and then "version", or the other way
around.  Whether or not the "version" column appears to the right or
the left of the "branch" column is irrelevant as far as usefulness is
concerned.  In non-"?expand"-view, it might be nice to allow "click on
category to see available versions" rather than "... to see available
branches", but as long as "?expand" is readily available it's no big
deal.)

To me, it does make a lot of sense to allow transposing version and
branch (in the what-to-sort-on sense).  I see the "branch" and
"version" as the labels of (orthogonal) the axes describing a
(per-category) plane.  Neither axis corresponds to a "finer scale"
than the other; the particular ordering found in Arch version names is
a result of a (nearly) arbitrary decision on which field should
precede the other in Arch version names.
-- 
Harald