[twg-tds] storing scripts in the texmf tree

[Paul Vojta]

On Sat, Feb 14, 2004 at 02:16:23PM +0100, Olaf Weber wrote:
> Paul Vojta writes:
> > I'm continually dismayed at the amount of effort that goes into opening
> > a few files.  (I'm one who likes short wide trees, not wide narrow ones.)
> 
> Of course the tds effectively mandates the latter, with all attendant
> problems, inorder to avoid having directories with thousands of
> entries.

I find that argument to be a bit circular, since the TDS grew out of kpse.

> Just to give an idea of how the trees are currently populated:
> 
> texmf-dist$ for i in *; do echo -n "$i        " && ls -R1 $i | wc; done
> aliases       1       1       8

[etc.]

On the other hand, without resorting to subdirectories, we have:

vojta at dell:~$ ls -R1 /usr/bin | wc
    849     849    7734
vojta at dell:~$ cd /usr/share/man
vojta at dell:/usr/share/man$ for i in *; do echo -n "$i        " && ls -R1 $i | wc; done
cs              6       5      43
de             22      19     193
es             83      79    1270
es_ES             13      11     133
fi              5       4      33
fr             78      74    1187
fr_FR              5       4      43
gl              5       4      33
hu             10       9      85
id             10       8      68
it             41      38     430
ja             77      74     974
ko             12      10      89
man1            860     860   11955
man2            304     304    4100
man3           2492    2492   45324
man4             40      40     460
man5            105     105    1490
man6              4       4      43
man7            112     112    1522
man8            271     271    3858
man9              2       2      21
pl             56      53     650
pt_BR             32      29     442
ru_RU             13      11     133
sv             25      22     262

This is on Debian linux, which relies on external databases to manage
the complexity of having ~2500 entries in /usr/share/man/man3, for
example.  This is a road that I would like to see kpse go down.

--Paul Vojta, vojta at math.berkeley.edu