[twg-tds] Re: [texhax] revision of the tex directory structure document

Reinhard Kotucha reinhard.kotucha at web.de
Sun May 30 03:07:07 CEST 2004

>>>>> "Karl" == Karl Berry <karl at freefriends.org> writes:

    >     What do you think about removing the "type/" directory?
    > Can't be done, much as I would like to.  That's where we
    > started.  Please see the "Font file type location" section.

    > http://tug.org/tds/#Font%20structure

It says:
 > but unless all the programs that search this tree employ some
 > form of caching, there are serious performance concerns.

I do not see any problems if the file is mentioned in an ls-R file.
In this case the ls-R file has to be read line-by-line and there is no
way to skip to the directory where the file is expected.

A severe performance problem will occur if the file is not mentioned
in an ls-R database.  Then it makes sense to restrict the search to
$TEXMF/fonts/type.  But on the other hand, kpathsea allows to control
which subtrees should be scanned on the disk and which ones should be
looked up in ls-R files.

It is reasonable to look into the ls-R files first and then into the
current directory (maybe recursive, but usually this tree is
relatively small).

Or are there any TDS compliant systems which do not have a database?

Does scanning the ls-R files influence performance?  It's quite
curious that "kpsewhich cmr10.tfm" needs 0.158 seconds and 
"kpsewhich -expand-path \$TEXMFFONTS" needs 0.151 seconds.

It seems that reading ls-R files is incredibly fast, texmf.cnf has to
be evaluated in both cases.

Does kpathsea call a shell to expand variables?


Reinhard Kotucha			              Phone: +49-511-4592165
Marschnerstr. 25
D-30167 Hannover	                      mailto:reinhard.kotucha at web.de
Microsoft isn't the answer. Microsoft is the question, and the answer is NO.

More information about the twg-tds mailing list