[Gimp-docs] Why do we profile 2 times

[Daniel Egger]

On 24.10.2006, at 22:03, Nickolay V. Shmyrev wrote:

>> Have you verified that this actually *does* work? The problem as
>> I remember it was for one, xincludes posed a problem because
>> xsltproc would not handle it in one pass as expected and the
>> second topic was the handling of the languages which caused
>> problems when done differently.

> Ups, sorry, I took some time for me to build gimp-help. Here are the
> results:
>
> Without patch
>
> real    16m19.951s
> user    10m46.960s
> sys     0m19.029s
>
> With patch
>
> real    12m9.448s
> user    10m20.895s
> sys     0m14.569s

> There is no difference between results, but simple diff failed to
> discover it. That's because of autogenerated id attributes. They  
> differ
> in different builds, that makes comparison harder.

There's no doubt it is faster to run less passes, however the real  
question
is whether this is equally correct concerning the final results in all
languages and output formats.

For instance it strikes me odd that the labels would be different; not
entirely improbable but strange anyway. Do the crosslinks still work?
Is the XML index file generated and correct?

It's good to hear it works for you and helps you easing a problem, but
before this goes in we really need to make sure it works for everyone
else as well. I can do this, but that'll take a while because it is
number 37 on my todo-list ATM.

Servus,
       Daniel

-------------- next part --------------
A non-text attachment was scrubbed...
Name: PGP.sig
Type: application/pgp-signature
Size: 186 bytes
Desc: This is a digitally signed message part
Url : /lists/gimp-docs/attachments/20061024/dc638111/PGP.bin