Thanks Tom/Ezra/Jeremy for your comments.
- How's possible that Links fails to appropriately compile the
queries? How can I detect that? In any case it fails I would receive
an error, isn't it?, but I haven't so far. During the development
process I continuously checked the error log (under logs/links-errors)
but it seems that everything's good so far.
- Ezra/Jeremy/Sam, the source code is in version 0.33, you can find it at
http://frege/~gabriel/linksPetStore/test33.links, is there any chance
you can take a quick look to see if you can detect something wrong?,
maybe give me some feedback on the responsiveness of the application.
It's supposed to be the final version.
- Tom, I've been through your report and I'm wondering how can I know
how much Javascript code is generated?, I ran some tests on the
wine.links application and the Average Page response time is 4.9 s,
quite good, probably because, as you said, there are not so much
access to the database.
- Jeremy, I'm working on frege at the moment, how can I turn on the
"measure_performance" flag?
Well, timing constraints for a summer project can be quite restrictive
(performance tests are not in the scope), however, having your
comments about my doubts and probably some feedback from the Links
team about the responsiveness of the application will be useful for
the future.
Thank you,
Gabriel
Quoting Jeremy Yallop <[EMAIL PROTECTED]>:
Quoting G Tellez <[EMAIL PROTECTED]>:
--> Average page response time is 8.64s.
--> Average page response time is 25.1s
These times are obviously pretty terrible. In part the slowness is
caused by the fact that the Links implementation currently performs
lexing, parsing, type-checking (twice!) and various analyses and
transformations to the program every time a page is loaded or a
client-server function call is made. Caching the result of these
phases should have a significant effect on performance, but we haven't
treated it as a priority so far. (I've just noticed that there is
actually a "cache_programs" setting which seems worth trying.)
It's not clear whether the difference between the times of the two
applications is caused by the fact that one simply takes longer to
type-check etc., or by Links generating inefficient queries, or by
something else altogether. It shouldn't be hard to find out, though.
You might start by setting the "measure_performance" flag and keeping
an eye on the debug output, besides looking at the generated SQL as
Ezra suggests.
Jeremy.
--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
_______________________________________________
links-users mailing list
[email protected]
http://lists.inf.ed.ac.uk/mailman/listinfo/links-users
