Re: [Gnu-arch-users] Debugging new user induction

[Davide Libenzi]

On Sat, 4 Oct 2003, Robert Collins wrote:

> What I think happens is that new users to arch have some more or less
> preconcieved ideas about their 'ideal' RCS system, and look to shoehorn
> arch into the same pattern. Then, when they are told there are better
> ways, the better ways are rejected - on the basis that the new user
> knows better.

(just when you think it's over)
Neither myself or Andrea ever claimed to "know better". As a matter of
fact, the more I learn in this world the more I discover I have holes in
my knowledge. There're a couple of things that I know pretty sure though.
(two issues myself and Andrea exposed were heavily criticized here, the
need of strict commits and the fact we didn't like the code to be littered
with tag-line: metadata)

1) Noone of the SCM I used before implemented the magic-drop[1] feature
inside the repository. Now, it is possible that everyone in the world ever
wanted this but no company/open-source-group ever implemented it. Or it is
possible that noone ever cared since the add/move/delete frequency is
ridiculously lower than commits. I'd exclude the fact that it was rocket
science so badly that noone was able to implement the thing. And because
of the universally adopted add/move/delete concept used by other SCMs, be
prepared to repeat threads like this in the future (at least if you
continue to blame people that requires it as default). An exception to this
(in my very personal view) is the initial import, where automatic tools of
selection might be useful.

2) Noone of the SCM I used before (even ones doing the right thing with
renames) ever suggested/required to blend metadata with data (tag-line:).
I used before two other SCMs that handles renames correctly (and I know BK
does even if I did not try it, so it makes three) and noone of those
require (like you suggest with tag-line:) your data to be littered to do
the right thing. Are those really genius like Larry repeatedly advertised
inside lkml ?

As I told many times before, I do not care if Arch has the tag-line:
(since I'll never use it in my sources) and I don't care about the
rule-guided commit (since I'll use precious untagged sources now). I
won't stop using Arch because of this. On the contrary, I'm trying to
introduce it in the company I work in. Bottom line, I'm glad Arch provides
me the feature I need (disconnected usage, change-set concept, etc...)
w/out forcing me to do things that I do not agree with.
So please stop bashing people that do think in a different way, when
"Previous Art" exist in this field.



- Davide


[1] magic-drop == non-strict-commits