[twg-tds] storing scripts in the texmf tree

Olaf Weber olaf at infovore.xs4all.nl
Sat Feb 14 17:05:42 CET 2004


Hans Hagen writes:
> At 21:43 13/02/2004, you wrote:

>> One possibility I'm exploring is using a 'texmf.zip' for the texmf
>> tree.  We have again the issue that (like building the hash table from
>> the ls-R file) building the index into the zip will be a fairly
>> expensive operation, which could conceivably benefit from things like
>> daemonizing or sharing.

> how complicated is it to let the tex family of programe transparently
> accept zip and gz files directly?

It's not _that_ hard, as the infrastructure is mostly in place in my
new code.  However, for individual files (e.g, a .gz'd file or a
single file in a .zip) it is unlikely to be of much use.  In a
filesystem, the granularity of storage is typically the filesystem
block size.  Filesystem block sizes are typically 1024 bytes (ext2),
2048 (ISO), 4096 (XFS), and let's not talk about (V)FAT.  Most of our
files are in this range of sizes, with very few larger.  This means
that compressing them singly doesn't gain storage, doesn't gain
improved I/O access times, and incurs a compression/decompression
penalty.  A net loss, in other words.

The exception is storing an entire tree into a zip file, where a large
part of the gains is not the compression but the fact that we do not
lose half a filesystem block per file.

-- 
Olaf Weber

               (This space left blank for technical reasons.)



More information about the twg-tds mailing list