perl script design question

[j proctor]

On the Jax.PM jacksonville-pm-list;
j proctor <jproctor at oit.umass.edu> wrote -



I think append and parse later is perfectly adequate for your needs this
time.  You're only talking about 1200 lines total; DBM may not actually be
that much faster (especially if you've got the overhead of generating
unique keys).

Depending on the client and web server software, you may find that it's
consistent in which order it reports the name/value pairs; I don't think
the HTTP standard requires this, but they're passed through the
environment (regardless of GET or POST) in a text string instead of a
hash, so it's probably fairly trustworthy.  Adding a record separator (a
line of ___'s or something) and knowledge of what order you're getting the
data can *greatly* speed up processing later.  Run some tests to be sure;
there may be some reason for a browser to deviate from orginial html form
order, but I can't think of it--if you were a browser that needed to send
a text string of form data back to a server, wouldn't you just go through
the file and append input field names and user-entered values?

If I had an actual database server running, I'd consider something like a
queued transactional update, but on a survey of this size it would really
be for the academic enjoyment of knowing I had written it than out of any
necessity.  


j



The Jacksonville Perl Monger's Group is operated by -
Bill -Sneex- Jones ( sneex at usa.net ),
to whom send all praises, complaints, or comments...