[twg-tds] storing scripts in the texmf tree

[Olaf Weber]

Paul Vojta writes:
> On Sat, Feb 14, 2004 at 03:56:30PM +0100, Hans Hagen wrote:
>> At 20:49 13/02/2004, Paul Vojta wrote:

>>>Of course, the logical next step would be a file system (kpsefs).

>>> :-)

>> actually, we should start thinking aloing those lines ... microsoft 
>> longhorn will have a database'd filesystem (sql + xml) so i expect the 
>> 'others' to follow up on that; in a way kpse is already a sort of 
>> intelligent way to access the tree, we'd better be prepared for that

> I draw the opposite conclusion.  If future OSes are going to implement
> kpse-like functionality, then we should wait for those features to appear,
> and make use of them.  The alternative would end up as (p)re-inventing
> the wheel.

Of course the future isn't here today yet.  Todays filesystems can be
viewed as databases that have specific keys and are strongly optimized
for handling that kind of key/data combination.  Whether the
characteristics of these new filesystems will be such that the search
functions of kpathsea will be (mostly) unnecessary when running on top
of them is at present not clear to me.

-- 
Olaf Weber

               (This space left blank for technical reasons.)