[Yanel-dev] should new pages' URLs end with ".xhtml" or ".html"?

Guillaume Déflache guillaume.deflache at wyona.com
Wed Apr 22 16:10:00 CEST 2009


Hi!

Michael Wechner schrieb:
> Guillaume Déflache schrieb:
>> Hi!
>>
>> When creating the new "Features" page ( 
>> http://yanel.org/en/features.html ) I saw that many pages in the Yanel 
>> website have both "*.xhtml" and "*.html" URLs(*)(**) (they use two 
>> entries in the repo's "paths" directory).
>>
>> Why is that? It looks like we had to do that at some point to be 
>> crawled by some tool that did not grok MIME-types but only use 
>> extensions: is that the reason?
> 
> IIRC Internet Explorer had some issues with xhtml suffix and xhtml mime 

Fair enough for the MIME-type as it has never been handled correctly so 
far by any IE AFAIK. See in particular 
http://bugzilla.wyona.com/cgi-bin/bugzilla/show_bug.cgi?id=4787
Do you remember any specific issue with the "xhtml" suffix? I doubt 
something gets triggered IE depending on an ".xhtml" extension given 
that it does not support XHTML at all, but you never know! ;) There may 
be issues with IE content sniffing though if we are serving XHTML 
content as HTML, but that's another problem.


> type and also people expect an html suffix.

IMHO "normal" people don't expect anything in that respect and just 
don't care, we as programmers *might* (now you know I don't! ;) ).


> But to make clear internally that this is really xhtml we added the 
> xhtml suffix.

I again don't think it's so important to know that here.
I'd rather have a meaningful extension in the content repo corresponding 
the content original MIME-type (but this would probably be a pain to 
implement that, so let's forget that! ;) )


Granted all this is pretty minor but I'd rather have to add two mappings 
each time I have to create a new page! :)


>> This may well be an issue if we want to move away from a VFS repo impl 
>> using path-mapping to something simpler for the website ( see 
>> http://bugzilla.wyona.com/cgi-bin/bugzilla/show_bug.cgi?id=6940 ) as 
>> then we would probably get two copies of the same content in "*.xhtml" 
>> and "*.html" files...
> 
> if we migrate to VFS, then we need to rename the xhtml to html, but 
> don't worry about this, because it will be done automatically anyway, 
> because only what the map contains is important

Sure, but the very problem is that the map itself contains both ".xhtml" 
and ".html" mappings, so I do not see how it is not going to generate 
duplicates, as would for example happen during a copy of directory with 
two symlinks pointing to the same file in a filesystem not "seeing" 
symbolic links but only what they point to (java.io.File API comes to 
mind...).
But anyway I probably don't get/remember how the repo import/export works.


>> (*): There are even some odder mappings, like both 
>> "acknowledgements.html.txt" and "acknowledgements.html" (plain text 
>> document).
> 
> this is, because it's based on Wiki syntax

OK.


>> (**): IMHO we should really use no extension at all for pages as we 
>> seem to use MIME-types correctly internally, but anyway...
> 
> I think extensions make a lot of sense still, because it gives a quick 
> clue what the mime type could be.

Well, I would argue that in that part of the repo we do not really care 
and it may be misleading and/or "incorrect", as the magic ultimately 
happens elsewhere e.g. in the resource-configs.

Cheers,
    Guillaume


More information about the Yanel-development mailing list