Re: [DOC-WEB] cvs: docweb /scripts notes_stats.php
[email protected] ("Nuno Lopes")
| Newsgroups | php.doc.web |
|---|---|
| Message-ID | <002a01c50a06$55d37490$0100a8c0@pc07653> |
> On Tuesday 01 February 2005 18:34, Nuno Lopes wrote: >> nlopess Tue Feb 1 12:34:44 2005 EDT >> >> Modified files: >> /docweb/scripts notes_stats.php >> Log: >> major script tuning: >> improved speed > > I don't know your machine, but on mine I only notice that it have to take > a 15 > minutes break for the script to finish the query (not counting the > download > time) > and the /www script seems to take 5 seconds instead of 0.25 seconds that > the > orignal version used (on my machine) > > Also the huge database it not needed at all, only wasting disk space > > ls -l notes_stats.sqlite* > -rw-r--r-- 1 Allowee Allowee 3975168 Feb 3 15:25 notes_stats.sqlite > -rw-r--r-- 1 Allowee Allowee 158720 Jan 31 22:22 > notes_stats.sqlite.old > > I have to agree that the old version could be improved even more, by only > inserting the top 20 files instead of all files is a good tweak.. > > Would be nice to hear why everything has to be changed while the orginal > version worked quite fine. My main machine is a Pentium4 2 Ghz with PHP 5.1. And the program is fast (although slower than the original). I think the original script was quite a mess, indentation broken,... And the most important thing: it had the logic broken. It splitted the notes by old/new in the cron job, which is wrong, because this is a comulative script. Over the time, older notes wouldn't be marked as such. What I've done is to create a database with all notes. This may seem a slow process for the first time, but following runs are faster. So, the presentation code shouldn't be in the cron script. Now we have two choices: generate the old DB schema, collecting data from the new one; or simply cache the results. The orphan notes cron job generates a static file. The same thing could be done with this script. Nuno P.S.: As usual, I don't mind to revert my changes