You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@forrest.apache.org by "Adam R. B. Jack" <aj...@trysybase.com> on 2004/05/25 21:37:45 UTC

No joy...

I'm trying the latest forrest (in batch mode and run) on Depot Version. (The
older copy of Forrest I had was old, and not producing PNGs from SVGs, so I
updated. FWIIW: I still don't get PNGs for the logo SVGs -- so I copied in a
converted one -- but that is the lest of the woes.)

Anyway:

1) I don't think Forrest likes directories called 'api'. Confuses the
bejeebez out of it.
2) I don't think Forrest likes pages called 'todo.xml' (even when not in the
root).
3) I don't think 'forrest' (batch) and 'forrest run' work comparably. I ran
both and got:
3.1) No menus on my batch pages
3.2) Different sized fonts
3.3) Different 'corners' (on tabs)
3.4) 'api' works with 'run', but not w/ batch. [I renamed to 'code', it
works now.]

I tried converting from book.xml to site.xml, and I admit I never understood
that thing (simple though it might be). Anyway, it seems that "Forrest run"
is happy, but "Forrest" is not. Lots of cocoon errors (about EOF), seeming

I'd put these into JIRA but I can't get past the NPE when I try to get
reminded of my password, and (frankly), I've burned more time than I
had/have & this is no fun what-so-ever. If you wish to experiment, please
look here for the site:

    http://svn.apache.org/repos/asf/incubator/depot/trunk/version/

Here is what I get from running forrest:

    http://cvs.apache.org/~ajack/x.txt

I know you are in the midst of a re-work, no complaints there -- stuff
happens -- but I really think forrest needs a regression test suite (that
includes some user sites). I just want a basic site, no frills, and I can't
get it working. I really wish that a shared/managed Apache Forrestbot
existed so perhaps you could see what I see. It'd make a darn good
regression test suite..

[BTW: The offer is still out there to integrate forrestbot (in ant or not)
into Gump, if Gump can checkout sites/run things. Whatever.]

regards,

Adam
--
Experience the Unwired Enterprise:
http://www.sybase.com/unwiredenterprise
Try Sybase: http://www.try.sybase.com


Re: No joy...

Posted by Juan Jose Pablos <ch...@che-che.com>.
Adam R. B. Jack escribió:
> I'm trying the latest forrest (in batch mode and run) on Depot Version. (The
> older copy of Forrest I had was old, and not producing PNGs from SVGs, so I
> updated. FWIIW: I still don't get PNGs for the logo SVGs -- so I copied in a
> converted one -- but that is the lest of the woes.)
> 

Hi Adam,

I had a look to your site, and first I tried to get it working under 
forrest 0.5.1, That version has support to produce png out of svg.

On src/documentation/content/xdocs/ant/sample_scripts.xml you have:
<!ENTITY files SYSTEM "files.ent">

But none of the files on that file seems to be available.


About the java docs an the junit:
http://xml.apache.org/forrest/faq.html#ignoring_javadocs

you need to ignore both so you do not see errors.

Let me know if you have more issues.

cheers,
Cheche

Re: No joy...

Posted by "Adam R. B. Jack" <aj...@trysybase.com>.
> 1) I don't think Forrest likes directories called 'api'. Confuses the
> bejeebez out of it.

     http://cvs.apache.org/~ajack/x-api.txt

This was with an 'api' subdirectory.

regards,

Adam

Re: Forrest (copyless-merged) bugs (was Re: No joy... )

Posted by Nicola Ken Barozzi <ni...@apache.org>.
Adam R. B. Jack wrote:
 > Nicola Ken Barozzi:
...
>>You mean that it does not transfer index.*xml* but it correctly
>>generates index.html?
>>
>>If so, this is correct and it has always done so. (a)
>>
>>If you don't get a correctly generated index.htmlwe have a problem. (b)
>>
>>>As this is an ant copy (I think), that makes some sense.
>>
>>Case (a) or (b)?
> 
> It used to be (a) with previous Forrest. It is currently (b) 'cos I was
> passing -Dproject.site-dir on the command line and the transfer/copy (of
> rss.xml) occurs to there, but the generation/transfer occurs to where
> project.site-dir is set in the forrest.property files. The things go to two
> places, not one. This is the command line -D issue you know exists. Any
> decision on if you'll fix it or not? 

I'll fix it(when I come back from my small vacation in Rimini next week :-).

> If not, I can deprecate Gump calling
> Forrest (might do so anyway) and only allow Gump to write xdocs into a
> Forrest webapp (nicer performance).

Probably you would want to do it in any case, so that it will "force" 
users to use 'forrest run' to see it locally and to install forrest 
webapp to see it on the website.

-- 
Nicola Ken Barozzi                   nicolaken@apache.org
             - verba volant, scripta manent -
    (discussions get forgotten, just code remains)
---------------------------------------------------------------------


Re: Forrest (copyless-merged) bugs (was Re: No joy... )

Posted by "Adam R. B. Jack" <aj...@trysybase.com>.
> Adam R. B. Jack wrote:
>
> > I put content that I do NOT ask Forrest to generate into ./content
> > [e.g. ./content/ant/rss.xml and ./content/ant/atom.xml] and the
> > content I DO ask Forrest to generate into ./content/xdocs [e.g.
> > ./content/xdocs/ant/index.xml.] Forrest transfers the things I
> > produce the rss.xml atom.xml are mine, not from Forrest) and it
> > generates index.html giving me:
> >
> > {site}/ant/rss.xml {site}/ant/atom.xml and {site}/ant/index.html
>
> Ok, so it *does* generate index.html.
>
> > What I am saying is the transfer occurs (I get the first two) but not
> > the last one.
>
> You mean that it does not transfer index.*xml* but it correctly
> generates index.html?
>
> If so, this is correct and it has always done so. (a)
>
> If you don't get a correctly generated index.htmlwe have a problem. (b)
>
> > As this is an ant copy (I think), that makes some sense.
>
> Case (a) or (b)?

It used to be (a) with previous Forrest. It is currently (b) 'cos I was
passing -Dproject.site-dir on the command line and the transfer/copy (of
rss.xml) occurs to there, but the generation/transfer occurs to where
project.site-dir is set in the forrest.property files. The things go to two
places, not one. This is the command line -D issue you know exists. Any
decision on if you'll fix it or not? If not, I can deprecate Gump calling
Forrest (might do so anyway) and only allow Gump to write xdocs into a
Forrest webapp (nicer performance).

regards,

Adam


Forrest (copyless-merged) bugs (was Re: No joy... )

Posted by Nicola Ken Barozzi <ni...@apache.org>.
Ok, we have our first detailed bug report after the copyless branch 
merge. Thanks, Adam! :-)

Adam R. B. Jack wrote:
 > 1) I don't think Forrest likes directories called 'api'. Confuses the
 > bejeebez out of it.

Ugh, this is not that nice. It shouold go away with locationmaps in the 
next releases, but for now renaming the dir or changing the cli xconf 
will do the trick.

...
>>>2) I don't think Forrest likes pages called 'todo.xml' (even when not in
>>> the root).
>>
>>You're right.  Delet the reference to it in sitemap.xmap to work around
>>that.
> 
> Could Forrest validate against such things, or give a nice warning/error
> message or something?

It should, yes. In fact yours is the first bug report for this, so we 
hadn't thought of it much before. Usually we have forrest to generate 
sites that users start from the template, not heavily pregenerated sites 
like Gump (that we *have* to support for DOCO).

What we should do:

- deprecate status.xml, todo.xml, etc handling by filename
- insert the recognition of them by DTD
- do a check in Ant that suggests to add a DTD to these files if
   it finds them

>>>3) I don't think 'forrest' (batch) and 'forrest run' work comparably. I
>>>ran
>>>both and got:
>>>3.1) No menus on my batch pages
>>>3.2) Different sized fonts
>>>3.3) Different 'corners' (on tabs)
>>
>>What skin?  I think some are currently broken.  todo: we need to check all
>>skins before a release candidate.
>
> Krysalis. I think it is likely me, due to my site.xml/tabs.xml or something,
> but I get a site with "forrest run" (as does somebody on same xdocs on older
> Forrest) but I get oddities w/ batch "forrest".

You mean that forrest run works ok and the batch one doesn't.
I know that the rounded corners are currently *not* generated with batch 
forrest - bug - but probably the last fix on the skins also broke the 
menus. Will have to check.

[Taken from Re: Build.sh and build.bat creating just JAR]

Adam R. B. Jack wrote:

> I put content that I do NOT ask Forrest to generate into ./content 
> [e.g. ./content/ant/rss.xml and ./content/ant/atom.xml] and the 
> content I DO ask Forrest to generate into ./content/xdocs [e.g. 
> ./content/xdocs/ant/index.xml.] Forrest transfers the things I 
> produce the rss.xml atom.xml are mine, not from Forrest) and it
> generates index.html giving me:
> 
> {site}/ant/rss.xml {site}/ant/atom.xml and {site}/ant/index.html

Ok, so it *does* generate index.html.

> What I am saying is the transfer occurs (I get the first two) but not
> the last one. 

You mean that it does not transfer index.*xml* but it correctly 
generates index.html?

If so, this is correct and it has always done so. (a)

If you don't get a correctly generated index.htmlwe have a problem. (b)

> As this is an ant copy (I think), that makes some sense.

Case (a) or (b)?

-- 
Nicola Ken Barozzi                   nicolaken@apache.org
             - verba volant, scripta manent -
    (discussions get forgotten, just code remains)
---------------------------------------------------------------------


Re: No joy...

Posted by "Adam R. B. Jack" <aj...@trysybase.com>.
> > 1) I don't think Forrest likes directories called 'api'. Confuses the
> > bejeebez out of it.
> The default cli.xconf excludes 'api' directories (the rationale involving
> links to javadocs).  Copy it from src/core/context/WEB-INF/cli.xconf to
> your project's conf dir (probably src/documentation/conf).  Then modify
> the <exclude>s.

Mine was an api/index.xml that discussed, then references .../javadoc/... So
I hit ot for exaclt that reason. What is bad (to this mere mortal) is the
copious error messages, non of which tell *me* anything useful like that.

Is it bad that I'd like to remain a dumb forrest user, and especially dumb
to Cocoon? I just want to generate documentation, I don't want to know what
is going on behind the scenes. I view Forrest as the tool built on Cocoon
for dumb users like me. If I am wrong/misguided, please let me now.

> > 2) I don't think Forrest likes pages called 'todo.xml' (even when not in
the
> > root).
>
> You're right.  Delet the reference to it in sitemap.xmap to work around
> that.

Could Forrest validate against such things, or give a nice warning/error
message or something?

> > 3) I don't think 'forrest' (batch) and 'forrest run' work comparably. I
ran
> > both and got:
> > 3.1) No menus on my batch pages
> > 3.2) Different sized fonts
> > 3.3) Different 'corners' (on tabs)
>
> What skin?  I think some are currently broken.  todo: we need to check all
> skins before a release candidate.
>

Krysalis. I think it is likely me, due to my site.xml/tabs.xml or something,
but I get a site with "forrest run" (as does somebody on same xdocs on older
Forrest) but I get oddities w/ batch "forrest".

> I can probably look into this later tonight... but I'm on company time for
> another hour :-)

Thanks.

> > get it working. I really wish that a shared/managed Apache Forrestbot
> > existed so perhaps you could see what I see. It'd make a darn good
> > regression test suite..
> >
>
> I agree...

Cool. Count me in. :-) Gump/Depot, I'd love to see their automatically
updated, I'd do twice as much documentation (or 10 times) if I could get it
built/updated easily (especially over my sad little modem).

> > [BTW: The offer is still out there to integrate forrestbot (in ant or
not)
> > into Gump, if Gump can checkout sites/run things. Whatever.]
>
> I think this has serious potential.  It's very new still, so we probably
> want to wait a few cycles and see it stablize before we can effectively
> find the commonalities with Gump.

Cool, whenever.

regards

Adam


Re: No joy...

Posted by Dave Brondsema <da...@brondsema.net>.
On Tue, 25 May 2004, Adam R. B. Jack wrote:

> I'm trying the latest forrest (in batch mode and run) on Depot Version. (The
> older copy of Forrest I had was old, and not producing PNGs from SVGs, so I
> updated. FWIIW: I still don't get PNGs for the logo SVGs -- so I copied in a
> converted one -- but that is the lest of the woes.)
>
> Anyway:
>
> 1) I don't think Forrest likes directories called 'api'. Confuses the
> bejeebez out of it.


The default cli.xconf excludes 'api' directories (the rationale involving
links to javadocs).  Copy it from src/core/context/WEB-INF/cli.xconf to
your project's conf dir (probably src/documentation/conf).  Then modify
the <exclude>s.

> 2) I don't think Forrest likes pages called 'todo.xml' (even when not in the
> root).

You're right.  Delet the reference to it in sitemap.xmap to work around
that.

> 3) I don't think 'forrest' (batch) and 'forrest run' work comparably. I ran
> both and got:
> 3.1) No menus on my batch pages
> 3.2) Different sized fonts
> 3.3) Different 'corners' (on tabs)

What skin?  I think some are currently broken.  todo: we need to check all
skins before a release candidate.

> 3.4) 'api' works with 'run', but not w/ batch. [I renamed to 'code', it
> works now.]

Right, the cli.xconf I mentioned is only used when the cocoon CLI
traverses (batch).

>
> I tried converting from book.xml to site.xml, and I admit I never understood
> that thing (simple though it might be). Anyway, it seems that "Forrest run"
> is happy, but "Forrest" is not. Lots of cocoon errors (about EOF), seeming
>
> I'd put these into JIRA but I can't get past the NPE when I try to get
> reminded of my password, and (frankly), I've burned more time than I
> had/have & this is no fun what-so-ever. If you wish to experiment, please
> look here for the site:
>
>     http://svn.apache.org/repos/asf/incubator/depot/trunk/version/
>

I can probably look into this later tonight... but I'm on company time for
another hour :-)

> Here is what I get from running forrest:
>
>     http://cvs.apache.org/~ajack/x.txt
>
> I know you are in the midst of a re-work, no complaints there -- stuff
> happens -- but I really think forrest needs a regression test suite (that
> includes some user sites). I just want a basic site, no frills, and I can't
> get it working. I really wish that a shared/managed Apache Forrestbot
> existed so perhaps you could see what I see. It'd make a darn good
> regression test suite..
>

I agree...

> [BTW: The offer is still out there to integrate forrestbot (in ant or not)
> into Gump, if Gump can checkout sites/run things. Whatever.]
>

I think this has serious potential.  It's very new still, so we probably
want to wait a few cycles and see it stablize before we can effectively
find the commonalities with Gump.  The forrestbot is basically a small
framework for automating getsrc/build/deploy/notify with Ant and various
ways to do each of those 4 parts.  The webapp is a pretty generic
authentication & execution front-end.  It could very easily execute
something besides forrestbot.

> regards,
>
> Adam
> --
> Experience the Unwired Enterprise:
> http://www.sybase.com/unwiredenterprise
> Try Sybase: http://www.try.sybase.com
>
>

-- 
Dave Brondsema : dave@brondsema.net
http://www.brondsema.net : personal
http://www.splike.com : programming
http://csx.calvin.edu : student org