Re: [Gnu-arch-users] Some issues

[Florian Weimer]

* Charles Duffy:

> On Wed, 2004-06-09 at 11:29, Florian Weimer wrote:
>> > I don't see how this is a design defect.  If you want to start fresh,
>> > use a new archive -- don't try to change history and deny the branch
>> > existed.
>> 
>> But this is exactly what creates stale branches. 8-)
>
> I think those are called "history".
>
> Storing history is generally considered a feature in a revision control
> system.

But usually, you also expect that the system provides some notion of a
HEAD branch.  Historic branches are just confusing.

> That's what branch sealing is good for.

Yes, I understand now that it's a solution for this problem.

> tla show-changeset is "additional tools" enough for this purpose. On a
> project where it's the primary revision control system, I hardly see how
> this is onerous (and, in the days of apt-get, emerge, etc, I hardly see
> how it's much of a problem in other projects either).

The absence of aepatch-style patches just makes tla a harder sell than
necessary, IMHO.  But I could be wrong.

>> I mean the checked-out trees, not the archive.  The archive format is
>> not such a pressing issue, indeed.
>
> ReiserFS is designed to promote just these kinds of "abuse" -- and the
> htree branch of ext3 isn't exactly slouchy either.

htree helps with large directories, but doesn't reduce wasted space in
files.

> There are filesystems built with performance in many-small-files
> configurations in mind; if this is, to you, such a pressing concern,
> you may wish to consider using one.

Most users (including me) run more applications than just tla, so tla
has to be efficiently on general-purpose file systems, too.