[tex-live] volunteer to implement lcab compression?

[Olaf Weber]

Thomas J Kacvinsky writes:

> Eeek!  Seeking around in a compressed file is not easy at all.

The cab works is that the files to be compressed are bundled into
"folders" -- concatenated, really.  These folders are compressed in
blocks of 32.  Among the data provided is where each file starts and
ends in the uncompressed data stream, and to which part of the
uncompressed stream each data block corresponds.  I think this means
you can skip to the right block, start decompression there, and drop
everything that comes before the data you want.  So random access has
additional costs, but they are not crippling.  And you keep them under
control by limiting the size of each "folder".

> I know that it would be relatively easy to implement ZIP compression
> in lcab, with the understanding that separate files go into separate
> data blocks (i.e., no one file is split between data blcoks), but I
> am not sure if the tools out there in *NIX land grok these types of
> cabinet files...

That would almost certainly be invalid under the CAB specs.

-- 
Olaf Weber

               (This space left blank for technical reasons.)