forrest-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Tim Williams" <william...@gmail.com>
Subject Re: FOR-1108 on Mac
Date Fri, 21 Nov 2008 03:51:08 GMT
On Thu, Nov 20, 2008 at 10:39 PM, David Crossley <crossley@apache.org> wrote:
> Tim Williams wrote:
>> David Crossley wrote:
>> > Tim Williams wrote:
>> >> Hey Devs-
>> >> I'm seeing FOR-1108 on OSX Leopard.  I would assume that JIRA is up to
>> >> date but that would mean trunk has been broken for quite some time, so
>> >> I thought I'd just ask before looking into it.  It seems to have been
>> >> reported only relative to a platform (windows) so I thought it weird
>> >> to be seeing it too.  Is the trunk still broken?
>> >>
>> >> https://issues.apache.org/jira/browse/FOR-1108
>> >
>> > As far as i remember, you are the first to report a
>> > problem on a modern Mac.
>> >
>> > Now searching the archives i found this from Sjur:
>> > http://marc.info/?l=forrest-dev&m=122112679627853
>> > "Re: [STATUS] [heads-up] merging our branch Cocoon-2.1"
>> > which indicates that it was okay on Leopard.
>> >
>> > Perhaps you can help to debug it, as it seems that
>> > our Windows developers have not yet found time.
>> > Cyriaque added a new observation recently to FOR-1108.
>> >
>> > Trunk works fine for me on this older Mac OSX Tiger.
>> > And it is fine on Solaris 10, as the forrestbot on the
>> > Forrest zones server is okay. Also fine for Thorsten
>> > and Brian on Linux.
>>
>> Actually, according to that mail, Sjur was only able to test in "run"
>> mode which does indeed work for me too - the problem occurs in forrest
>> site.
>
> What are the errors that you are seeing?
>
>>   There is actually a note at the bottom of
>> /samples-b/linking.html that gives a obscure clue - reading that note
>> it's still not clear to me why it fails to harshly at this point.
>
> That note is attempting to explain that the generated
> docs (as deployed on the zone) are not at the root of
> the webserver, hence the generated paths for some of
> the image references are not showing correctly.
> I don't understand how that would relate to FOR-1108.

Ok, apologies in advance if I'm missing something incredibly obvious -
I'm coming from a long hibernation...

So, I'm failing with a NPE just like the attached log file in
FOR-1108.  Looking into this led me to linking.html.  Specifically,
these output lines:

* [88/12]   [12/50]   0.155s 30.3Kb  samples-b/linking.html
* [97/3]    [0/0]     0.09s  1.9Kb   /images/ellipse-2.png
ERROR - Image not found: cocoon:/images/icon-b.png
ERROR - Image not found: cocoon:/images/icon-b.png
java.lang.NullPointerException
	at org.apache.cocoon.environment.AbstractEnvironment.release(AbstractEnvironment.java:563)
	at org.apache.cocoon.environment.wrapper.MutableEnvironmentFacade.release(MutableEnvironmentFacade.java:314)
	at org.apache.cocoon.generation.FileGenerator.recycle(FileGenerator.java:64)


So, I think, icon-b.png - let me investigate that a bit.  I notice the
comments and they don't immediately make sense.  Then I look at my
file system and, sure enough, forrest has placed icon-b.png in an
images directory that is sibling to the site build directory. This is
why I think they're related.  Then... I finally read the comments on
the reference to 'icon-b.png':

"Broken relative reference (it points up beyond the document root).
Never mind, Forrest "absolutizes" the URLs to be like the URLs in
examples 1 and 2 and 3. The source file is found in
src/documentation/resources/images/ directory. However there is a
problem here. When using Forrest in command-line mode, the image will
be generated outside the document directory. It is better to use
absolute /image/ URLs or be very careful about using such relative
links (Forrest will not report an error)."

It's starting to make a little more sense, but while I'm trying to
figure out what is exactly breaking, I keeping wondering why in the
heck we want to even to try overcome a site admin being a complete
moron in the first place?  Throwing an NPE is bad, but I'm thinking
trying to be this helpful in the first place is also bad.  But then
again, I've been away for quite some time, so this could eventually
turn out to be a user error :<\

--tim

Mime
View raw message