Re[2]: [PHP-DOC] Critical files in translation

2005-10-09 Thread anatoly techtonik
||*()*|| Hi, Sean.

>> P.S. I better prefer approach when outdated status is detected by
>> number of lines changed between revisions. I have some sketches for
>> PHP CVS protocol which can be used to do that without shell access.
>> Perhaps this information can be cashed in livedocs. Livedoc authors
>> can also find it convenient to flag critical revisions manually.

SC> One thing I think we should be careful of is resources.. both on the CVS
SC> server and on the Livedocs building machine.

That's right, but I thougth livedocs work on the fly and use cache
heavily, don't they? The amount of lines changed can be cached and
incrementally calculated without additional CVS access.

SC> One of the main reasons for introducing Livedocs is because the build of
SC> the english HTML manual takes >5 hours (on the production box.. last I
SC> heard).

Offtopic warning.
I've stopped work on XSL templates, because on my machine it took
about full working day to rebuild manual and resulting CHM file is
broken despite of what it successfully compiled. Seems like these
are bugs of indexing machine inside MS compiler, but I can't do
anything about it and haven't found a workaround yet. To test it
further I can only say what 256Mb box in not enough. =) I think
optimal configuration for building machine is about 1-2Mb. Better
1.5-2Mb.

SC> We _can't_ get into that situation ever again.

SC> (What I'm saying is diffing against english is expensive.. relatively
SC> speaking)

With incremental stats it is ok. Wiki holds changes/diffs in DB, and
livedocs can do the same.

t
-- 


Re[2]: [PHP-DOC] Critical files in translation

2005-10-06 Thread anatoly techtonik
||*()*|| Hi, Gabor.

>>>Now Sean, Derick, how does the automatic livedocs package rsync
>>>project is, so we can step forward with livedocs?
>> 
>> As always.. busy.
>> This week, however, I'm TOO busy.
>> The earliest I can look at this again is next week.

GH> The nice thing is that once this is set up, we only need to find people
GH> to handle the flow of bug reports we are going to get for livedocs,
GH> otherwise the packaging will work automatically :)

The biggest problem with these people would be to explain them
livedocs magic. =)

P.S. I better prefer approach when outdated status is detected by
number of lines changed between revisions. I have some sketches for
PHP CVS protocol which can be used to do that without shell access.
Perhaps this information can be cashed in livedocs. Livedoc authors
can also find it convenient to flag critical revisions manually.


t
--