Re: AS Libraries for stay-open CGIs
Re: AS Libraries for stay-open CGIs
- Subject: Re: AS Libraries for stay-open CGIs
- From: SeaSoft Systems <email@hidden>
- Date: Fri, 16 Feb 2001 12:51:41 -0800
Thank you, Jeff, for a most useful discussion!
At 14:05 -0500 2/16/2001, email@hidden wrote:
May I ask what these scripts do? If it's a shopping cart system, we must
trade notes.
No, it is a pretty specialized environment. Users log onto their
server realm/directory and the CGI's do various tasks for them (such
as simple directory management tasks, execution of server-based
engineering applications, etc.)
I have implemented exactly what you are after. Here are my insights.
1) You do not want "a single stay-open library of utility handlers accessible
to all the CGIs." This will choke under load because it would not be
multithreaded, and my understanding is that it would use a FILO (same as
LIFO, but the acronym order highlights the problem) system for responding to
requests, leading to slowness or worse, timeouts.
Wow. This is a biggie I hadn't considered...
One thing I am unclear on. Does the single-threaded nature of script
objects apply to an entire "script object library".
That is, if I have two separate stay-open CGIs, (presumably) they can
be in "simultaneous" play, each by separate users (right?). If I then
combine the two into a *single* stay-open CGI script with two entry
points, does the single-threaded nature of AS *prevent* simultaneous
execution of the two handlers because they reside in a single
stay-open script? Since my handlers can take some time to complete
their tasks (some being finder-related file operations), simultaneous
execution is important here.
4) You do want to load the script object library into a property retained
between runs as I demonstrated in a recent post (email me if you missed that).
Sorry, but I did miss that; I would be most grateful to see it.
6) Optimizing Applescript for speed is a valuable learning experience. I now
have a vanilla handler which is 4 times faster than an osax.
Glad to see serious CGI discussion happening.
Likewise...
Thanks again for the insights!
Richard Hartman