Re: Localization, Internationalization, Multilanguatisation

Wim Niemans ri wrote:
> I wish truly that you guys stop confusing me with mtime() and fmtime().
> They don't exist. Probably you mean filemtime() in favor of filectime() 
> ???

SOrry. mtime means "modification time". Some Linux tools have this as an 
option (find for instance). I didn't check the php name (I won't tell 
you  here what I think of php function names :-)

> Let me tell you the following:
> 
> All those stats functions are cached by PHP. So you don't have any means 
> to detect in production that a file has modified.

I remember reading that there's a way to flush changes and force reading 
file properties.

> Also, if a file is deleted you won't detect it.

Strongly doubts here.

> Finally, if I 'move' a file from a backup into the tree you would gladly 
> ignore that.
> 
> So, what do you mean exactly with 'caching' ?
> And what will be cached: scripts, templates, confs, output pages ?
> I don't have a clue and I'm feeling totally lost.

Many things.
Every slow operation that would be repeated for many pages before 
producing a different result should have this result cached.

If conf is long to load all values, conf should be cached.
Since lang strings will always change slower than php/tpl/ndf files will 
be used, files with lang strings replaced should be cached. This is the 
case with our current phing-time localization. I propose to keep this 
behaviour but wait till someone asks for a page to rebuild it.

And if output pages don't vary, they can -- already -- be cached.
-- 
Jean-Christophe Michel