David Crossley wrote:
Gav.... wrote:

From: Ross Gardler

David Crossley (JIRA) wrote:

Thanks. I am on UNIX and can see the problem line-endings in your
Windows-generated attachment, e.g. at lines 38 and 39 ...

    35 </head>^M$
    36 <body>^M$
    37 <div id="container">^M$
    38 <!--+$
    39     | start breadtrail$
    40     +-->^M$
    41 <div class="breadtrail">^M$

I already did a thorought analysis oif this problem - see the linked
messages. The above highlights the problem - it is *only* in the
comments that are the problem. This is confirmed in the Xalan issue this
one depends on.

Yes, I guess David wanted another confirmation from another Windows user.


Ross' Description of FOR-492 is back-to-front in its
explanation of line-endings. Hence my request on 04/Mar
for an explicit example. We need to change that Description.

Ahhh... now I understand the confusion. Sorry.

Anyway, seeing as the XALANJ-656 issue has not been updated since January,
has Anyone tried your suggestion of using FixCRLF in Ant ?

Do you want me to try it on my system?


I don't see how that will work, as it would just
mangle the rest of the line-endings for Windows users.

It is possible to force all line endings to be the same. SVN will then do the rest. The problem is, if I understand it correctly, the inconsitent line endings confuse SVN. Consistent line endings can be handled correctly.

I used this technique somewhere in the plugin system (can't remember where now).

but...

It would be better to follow up on the XALAN issue.
I have wanted to for a long time but i couldn't get
anyone to comment on my FOR-492 query. Now we can
point them to further information.

Yes, we should really address the root cause. That is why I never implemented the Ant workaround, and I guess why others have not done so.

It really should be something simple. It's just a case of finding the time to find the relevant part in the Xalan code. However, since it is hitting more and more people perhaps we should put a well commented workaround in place.

(in other words, Gav it would be worth the experiment if you want to do that rather than dig into Xalan)

Ross