Gentlemen,
I've just committed to CVS the new implementation of the MostWanted page
based on the new tables 'linked' and 'unlinked'. I've left caching out but
it still takes quite a while to compute. If this turns out to be a problem I
will program it back in.
-- Jan Hidders
On mar, 2002-02-19 at 16:26, Jimmy Wales wrote:
> Brion Vibber wrote:
> > Jimbo, is there any chance we can move the Esperanto wikipedia over to
> > the PHP script soon? I've been promising people we'd be upgrading to the
> > new software (which will fix a number of annoying bugs in the old) for a
> > while, and the natives are getting restless. :)
>
> Yes!
>
> > I'm going to check in a couple more character set and case-conversion
> > fixes tonight, after which we should be ready anytime. At this point any
> > additional problems are only going to be discovered by having real users
> > bang at the real site with real non-English non-ISO-8859-1 text...
>
> How about this -- tomorrow morning, I will install this. Will you be around
> (in email) tomorrow for questions?
Sounds great! If I remember correctly, we are both on Pacific time
(UTC-8), yes?
> Best thing to do -- send me simple step by step instructions,
> including instructions about the conversion script. I'll back
> everything up, run the conversion, install the new software, and
> it'll work perfectly the first try! (ha ha!)
It's relatively simple (famous last words). For everybody's reference:
1. Edit convertWiki2SQL.php to set some options there. Right now it's a
little rough, eventually it may or may not get smoother. Basically,
uncomment the special settings for the target language, and set the
$rootDir variable to point to the "page" subdirectory of the usemod db
that the data is being sucked from.
2. Run "php convertWiki2SQL.php" as a user with write permission to the
wiki directory. This should spit out a bunch of article titles and
create a big file called "newiki.sql" which contains the SQL commands to
insert everything into the database. Note that it's normal to see a few
errors about not being able to open a directory -- that just means
there's a letter of the alphabet that no article titles start with.
3. Create the database. I've been calling my test database "wikieo", but
whatever sounds appropriate should work just as well. Something like:
mysql -e "create database wikieo;"
should do it if I recall correctly. You might also have to specify the
proper username and password, I don't know how you guys have mysql set
up.
4. Initialize the tables and enter the data. Something like:
mysql wikieo < wikipedia.sql
mysql wikieo < newiki.sql
(After this step "newiki.sql" shouldn't be necessary anymore.) I'm also
going to put together a file with SQL commands to fix some articles with
extra uppercase letters in the titles which you can run here:
mysql wikieo < titlefix.sql
5. Edit wikiLocalSettings.php to set the language and database name.
Roughly:
$wikiLanguage = "eo" ;
$wikiSQLServer = "wikieo" ;
and if the hostname isn't being automatically picked up:
$wikiCurrentServer = "http://eo.wikipedia.com" ;
In theory everything should automagically work after that... (Assuming
of course that the apache rewrite rules are set up properly, etc.)
> If it isn't working perfectly right out of the box, then I'll back out
> the change, revert to the Usemod script, and we'll do a dry run in a
> "safer" way with test.wikipedia.com or whatever.
Great, I'll warn the others. :)
-- brion vibber (brion @ pobox.com)
Brion Vibber wrote:
> Jimbo, is there any chance we can move the Esperanto wikipedia over to
> the PHP script soon? I've been promising people we'd be upgrading to the
> new software (which will fix a number of annoying bugs in the old) for a
> while, and the natives are getting restless. :)
Yes!
> I'm going to check in a couple more character set and case-conversion
> fixes tonight, after which we should be ready anytime. At this point any
> additional problems are only going to be discovered by having real users
> bang at the real site with real non-English non-ISO-8859-1 text...
How about this -- tomorrow morning, I will install this. Will you be around
(in email) tomorrow for questions?
Best thing to do -- send me simple step by step instructions,
including instructions about the conversion script. I'll back
everything up, run the conversion, install the new software, and
it'll work perfectly the first try! (ha ha!)
If it isn't working perfectly right out of the box, then I'll back out
the change, revert to the Usemod script, and we'll do a dry run in a
"safer" way with test.wikipedia.com or whatever.
> There are a few switches at the top of convertWiki2SQL.php for selecting
> language-specific processing options, and wikiLocalSettings.php needs to
> select the proper $wikiLanguage and $wikiSQLServer, but that's about it.
> Other than those two, the PHP source files can be shared 100% between
> various language wikipedias. (I've already included the Esperanto
> message-localization file in the CVS repository, and I assume others
> will be added as they are converted & translated.)
Cool! I don't know anything about the conversion script, though. Jason did it
for the main site, and he's out of town. Oh, wait, he'll be back tomorrow. But
still, if you have information, let me know. :-)
----- End forwarded message -----
Jimbo, is there any chance we can move the Esperanto wikipedia over to
the PHP script soon? I've been promising people we'd be upgrading to the
new software (which will fix a number of annoying bugs in the old) for a
while, and the natives are getting restless. :)
I'm going to check in a couple more character set and case-conversion
fixes tonight, after which we should be ready anytime. At this point any
additional problems are only going to be discovered by having real users
bang at the real site with real non-English non-ISO-8859-1 text...
There are a few switches at the top of convertWiki2SQL.php for selecting
language-specific processing options, and wikiLocalSettings.php needs to
select the proper $wikiLanguage and $wikiSQLServer, but that's about it.
Other than those two, the PHP source files can be shared 100% between
various language wikipedias. (I've already included the Esperanto
message-localization file in the CVS repository, and I assume others
will be added as they are converted & translated.)
-- brion vibber (brion @ pobox.com)
Dear fellow programmers,
I have extended the database schema with two new tables: 'linked' and
'unlinked'. As usual the SQL for this addition can be found in
updSchema.sql. However, in this case the contents of the tables cannot be
generated by SQL alone, so there is an extra upLinks.php script in PHP that
contains the PHP code to do so. Read this file for further instructions.
The intention of these tables is to replace the 'cur_linked_links' and
'cur_unlinked_links' columns in the cur table. This will make it possible to
give the special pages that query linking information reasonable response
times, so they won't have to be cached. Right now I've only added the code
to keep these tables up-to-date and they are not used yet. From this moment
on the usage of the 'cur_linked_links' and 'cur_unlinked_links' columns is
depreciated, and I will remove them as soon as all the code that uses them
has been replaced by code that uses the new tables.
-- Jan Hidders
What versions of apache, php and mysql is everybody using? I want to
include the most appropriate versions in the README.
Myself:
# ../bin/mysqld --version
../bin/mysqld Ver 4.0.0-alpha for pc-linux-gnu on i686
http://127.0.0.1/server-status
Server Version: Apache/1.3.20 (Unix) PHP/4.0.6
I suspect that my intermittent database problems have to do with the
4.0.0 mysql and its handling of fulltext indices. I'll downgrade
shortly.
Axel
Hi,
I started my mysqld with --log and found that we are constantly and
pointlessly selecting the wiki database. I combined all those calls in
getDBconnection() and made sure that we only call mysql_select_db()
and mysql_pconnect() when we really need to (i.e. once), using a new
global $wikiDBconnection.
Axel
For a few glorious days, I had sysop status. I no longer do. Magnus
knows at least a little about the problem, but if any of you know as well
and can fix the problem, that would be great.
The problem might be related to a login problem we still have. When I go
to any page and press "Log out" and then "Log in", I come to a page that
says: "User Larry_Sanger1, you are already logged in!"
If I then attempt to log in as "Larry Sanger", which I *suspect* (don't
know) has sysop status (whereas "Larry_Sanger" does not), it tells me:
"Larry Sanger, you are logged in!" Moreover, the link in the upper right
hand corner is to "Larry Sanger"'s homepage. Hopefully, I then go back to
the Main Page only to find that I am once again "Larry_Sanger" (so says
the upper right corner link) and do not have sysop privileges.
I am running the current CVS version locally and updated my database
scheme. Search, RecentChanges, edit and preview works fine, but if I
try to save my changes, they are all lost and the file is back to the
previous version. It doesn't happen for all files, and I don't know
what the difference is (two example problem files are the main page
and "British Queen"). It's not a browser cache issue, since when I try
to edit the page again, the wiki text is also reverted to the old
version. I can write to the database, since the page counters change
correctly. Does anybody see that problem?
Axel