forrest-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ross Gardler <rgard...@apache.org>
Subject Re: [jira] Commented: (FOR-635) images not reproduced in PDFs, if sources are in xdocs/images directory
Date Sat, 13 May 2006 20:57:50 GMT
Gav.... wrote:
> 
>>-----Original Message-----
>>From: Ross Gardler (JIRA) [mailto:jira@apache.org]
>>Sent: Saturday, 13 May 2006 9:29 AM
>>To: dev@forrest.apache.org
>>Subject: [jira] Commented: (FOR-635) images not reproduced in PDFs, if
>>sources are in xdocs/images directory
>>
>>    [ http://issues.apache.org/jira/browse/FOR-
>>635?page=comments#action_12383339 ]
>>
>>Ross Gardler commented on FOR-635:
>>----------------------------------
>>
>>(some notes I'm working on about 3 issues at once due to Forrest Friday
>>and IRC - need this so we don't forget what we have discovered)
>>
>>Looking at http://localhost:8888/samples/sample.fo
>>
>>we see that an xdoc entry of:
>>
>><icon height="22" width="26" src="../images/icon.png" alt="feather"/>
> 
> 
> Why use ../ at all ?

It's legacy, but that is not important, the key is that the URL works 
for HTML but not for PDF. We need to standardise somewhere. Sure it 
should be in the sources, but then we can't control the users sources. 
If we can make the handling of URLs work regardless of the rubbish put 
in by users then Forrest is a big hit with users.

> The two images directories are static so this relevant from document rubbish
> Is not needed. 
> 
> And, is ../images intended to be /resources/images or /xdocs/images ?

Again, legacy. Images used to be in one location, then moved to another, 
then to another etc.

> There can also be ../../../images/icon.png if the nested directories are
> deep enough. In all nested or root cases specifying /images/icon.png would
> Always work. Can we do away with dotdots ?

Only if you are prepared to edit every single users source documents for 
them ;-)

Seriously, we can choose to only support images/icon.png and correct our 
sources. But if we can it would be better if we handle all potential 
user error - less work for us on the user lists.

So how do we do it?

How about stripping ay leading dotdots before we process image references?


(NOTE we should still tidy up our example docs to conform to our "best 
practice" recomendations)

I don't think the idea of adding special matches for possible matches is 
a good idea, that could be a never ending task.


Ross

Mime
View raw message