Hugh Gibson wrote:
> Just to clarify this point: I suggest that you specify a small timeout
> e.g. 1 minute. Obviously the viewer caches data files itself in that they
> are only loaded as required. Therefore the small timeout has no effect
> during a single session of using the API viewer. However, if the user
> navigates away then back to the API viewer, the browser will check with
> the server whether each data file (as well as the main script!) is still
> valid by sending the local date/time. The server checks if the file has
> been modified since then, and if not, sends back the "304 Not Modified"
> response. The browser can then use the local copy. This dramatically
> reduces the load on the server, and also ensures that any user will be
> using the latest data files within a very short time of them being
> deployed.
>   

Cache headers and compression are now implemented on the
demo.qooxdoo.org web server.

- The cache header returned by the web server look good to me. I set a
default expiration of 1 day, and the suggested 1 minute for Javascript
files.

- My tests (with FF) suggest that there is really no harm in such a
short expiration period. This browser even makes requests *before* the
60 seconds are over, just to receive the 304 response code. I think
that's not an issue for the web server. Nevertheless, I'd be happy if
people keep an eye on that and report back any issue they might encounter.

- On-the-fly compression is enabled with mod_deflate. As far as I can
see it looks good, but I'm concerned about the 'old browser' issue. I
have none of those (ugly) additional Apache config directives in place
of the kind  "BrowserMatch ...", since the heuristics look fairly
fragile to me. So currently mod_deflate just uses the 'Accept-Encoding'
request header to decide whether to compress or not (I believe). If this
brings up any kind of problem please let me know and I commit to the
BrowserMatch thingies.

- I'm also interested on qualified feedback on the over-all performance
in the api viewer under the new scheme, especially when slow lines are
involved.

> Sorry if this is teaching my grandma to suck eggs - but I feel that you
> are missing a trick here.
>
>   
I for my part are quite happy that you made that point, and on that
level of detail ;-). Also thanks to the other people on the list who
provided all the good links and details. You saved me hours of research.

Cheers,
=Thomas


-------------------------------------------------------------------------
This SF.net email is sponsored by DB2 Express
Download DB2 Express C - the FREE version of DB2 express and take
control of your XML. No limits. Just data. Click to get it now.
http://sourceforge.net/powerbar/db2/
_______________________________________________
qooxdoo-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/qooxdoo-devel

Reply via email to