Tags, being short strings commonly used, compress really nicely with
the common compression algorithms out there.
I once did an experiment (to confirm one that Gavin Nicol had done)
using a document with several thousand lines, each with one start-tag,
end-tag pair and no content. I tried compressing this file with
* no minimization
* short end-tags
* end-tag ommission
The uncompressed file was something like 50K. The compressed files
differed by only a few 100bytes from each other. The gains from short
end-tags did not carry over into the compressed versions, and the
compressed versions were so much smaller there seemed little
contest.
Having short-tag ommision can only compress a document by less
than 50% at an improbable maximum (in the case of a document with
not data, not attributes, no white-space, and incredibly long GIs).
Compression is a far better approach.
Rick Jelliffe