XDocReport, IDE support (Was: Re: Anybody with deeper OSGi bundle can comment on freemarker.jar OSGification?)

Monday, June 3, 2013, 11:16:59 AM, Angelo zerr wrote:

> Hi Daniel
>
>> Sure. You should provide a project description, less than 250
>>characters.
>
> Ok I will send you that.
>
>>Any ideas that would make FreeMarker a better fit for these kind of
>>templates that you enter right into Word/LO?
> docx/odt are a zip which contains several XML entries (ex: for docx
> word/document.xml, for odt content.xml)
> XDocReport unzip docx/odt and load some XML entries (
> word/document.xml, for odt content.xml) with Freemarker.
> So we have a Template instance for some XML entries. When report
> must be generated, XDocReport uses Freemarker to merge
> XML entries with Java context and create a new zip with XML entries merged.
> Freemarker works great.

I mostly meant issues with FreeMarker that has effect on the end-user.
Like if FreeMarker could to this-and-that, then you could add
this-and-that extra feature, or the end user wouldn't run into certain
error repeatedly, or at least would understand easier what's going
on... etc.

For example, what happens if the users enters `x && y`? Won't Word/OO
corrupt that to `x && y`?

> The only thing that I don't like is our extractor fields feature
> (it extract Freemarker fields name from docx).See method
> extractVariables
> https://code.google.com/p/xdocreport/source/browse/template/fr.opensagres.xdocreport.template.freemarker/src/main/java/fr/opensagres/xdocreport/template/freemarker/FreemarkerTemplateEngine.java
>
> I find this code is awfull:
>
> if ( DOLLAR_VARIABLE.equals(
> templateElement.getClass().getSimpleName() ) )

You mean, the problem here is that the AST is not (fully) exposed
through a public API. Well... once it's exposed, we have backward
compatibility constraints there too (and BC is already a very
problematic roadblock for FM). So at the very least the parser had to
be cleaned up before that. (Or maybe, it should remain internal, but
with some public tooling API that lets you traverse a higher level
view of the template, hiding some of the implementation details.
Although keeping that up to date is a lot fiddling in itself.)

Otherwise I don't 100% get what that part does. You extract the the
expression from inside the ${...} and make it the name of a filed. But
if I have `${1 + x}` then will the field name will be "1 + x", which
is maybe unintended.

And does the user really have to create a field to enter an
interpolation, or can he just enter ${expression} anywhere? (The point
of ${expression} instead of the $varName of Velocity is that ${...} is
a safe enough delimiter in itself in most applications.)

>> But it's on the Central Maven Repository for a long time. Or what do
>>you mean?
> yes you are right, but it doesn't support OSGi.

Of course, starting from 2.3.20, the Maven artifact will be an OSGi
bundle.

>> Yes, better tooling would mean a lot to FreeMarker.
> cool.
>
>> Was that also a problem with FM 3?
> Jonathan Revusky told me that it's possible to manage tolerant
> parser with FreeCC, but he has gaven Freemarker.
> I have not time to study FreeCC and I would like find people who
> are interested to help me to develop DLTK Freemarker.

He has also gave up on FreeCC as far as I know. But I suppose it can
be implemented with JavaCC too.

> An other idea that I had had, is to manage Freemarker with
> CodeMirro( Javascript editor which manages any mode). I have works a
> lot with CodeMirror for completion and syntax error.
> For instance you can find sample with Velocity
> http://codemirror-java.opensagres.cloudbees.net/velocity.html the
> syntax error is managed on server side. We could do the same thing
> with Freemarker and better manage debugging feature.

BTW, does Velocity have an error-tolerant AST builder?

> Regards Angelo
>
>
>
>
>
> 2013/6/3 Daniel Dekany <[email protected]>
> Monday, June 3, 2013, 1:21:04 AM, Angelo zerr wrote:
>
>> Hi everybody,
>>
>> @Daniel : for OSGi topic, it seems your bnd is good. We use
>> Freemarker with XDocReport which is Java reporting tool (you design
>> your report with MS Word or OpenOffice and you use
>> Velocity/Freemarker syntax to replace simple, list value) and
>> generate docx or odt report (and convert it with PDF or XHTML)
>> (Daniel, you can add XDocReport to the Powered By if you are agree,
>> thank's!).
>
> Sure. You should provide a project description, less than 250
> characters.
>
> Any ideas that would make FreeMarker a better fit for these kind of
> templates that you enter right into Word/LO?
>
>> We have an Eclipse RCP/RAP demo (which works on OSGi
>> context) and in this demo we use the com.springsource.freemarker
>> bundles and the MANIFEST.MF looks like your bnd.
>>
>> It's out of scope of this post, but I think it should really cool if
>> Freemarker could be hosted on the central maven repository (like
>> springsource have done). Many project uses Maven (we use it in
>> XDocReport) and I could simply switch to the "official" Freemarker
>> dependency.
>
> But it's on the Central Maven Repository for a long time. Or what do
> you mean?
>
> <dependency>
>   <groupId>org.freemarker</groupId>
>   <artifactId>freemarker</artifactId>
>   <version>...</version>
> </dependency>
>
>> @Debugging feature : I think today Freemarker is very mature and
>> stable, but I think tooling about Freemarker misses.
>
> Yes, better tooling would mean a lot to FreeMarker. I was rather
> unhappy about all that happened with your work on the DLTK plugin due
> to the cancelling of FM3.
>
>> It exists a JBoss Freemarker plugins but I think we can do better
>> than this plugin like managing debugger, preview, etc. I had started
>> this work with Eclipse DLTK at
>> http://sourceforge.net/p/freemarker/code/HEAD/tree/sandbox/org.eclipse.dltk.freemarker/
>> You can see doc at
>> http://sourceforge.net/p/freemarker/code/HEAD/tree/sandbox/org.eclipse.dltk.freemarker/docs/
>> and for debug I have used the Debugger Freemarker to manage DBGP protocol.
>>
>> But I have gaven my work because of the Freemarker parser which is
>> not tolerant (it crashes as soon as it  finds an error). With editor
>> context, parser must be tolerant (to build an AST even if there are errors).
>
> Was that also a problem with FM 3? Anyway, if you still want to work
> on that plugin, we could revive this topic, and then I could do
> something about it. Something that this time makes into a release.
>
>> IMHO, I think Freemarker should be improved  for tools support like :
>>
>>  * manage tolerant parser.
>>  * improve debugger (like DBGP debugger).
>>
>> Regards Angelo
>
> --
> Thanks,
>  Daniel Dekany
>
>

-- 
Thanks,
 Daniel Dekany


------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite
It's a free troubleshooting tool designed for production
Get down to code-level detail for bottlenecks, with <2% overhead.
Download for free and get started troubleshooting in minutes.
http://p.sf.net/sfu/appdyn_d2d_ap2