On more thing... the hard part in an evaluator tool is defining the
data-model. On freemarker-online that's a very limited due to security
reasons. In a tool that the user runs on his on computer though, it's
not a problem to allow defining values in Groovy for example.


Saturday, January 7, 2017, 6:45:10 PM, Daniel Dekany wrote:

> Saturday, January 7, 2017, 4:37:12 PM, Denis Bredelet wrote:
>
> [snip]
>>> These are some getting started examples. What are your ideas about the
>>> final product?
>>
>> I am thinking of adding a « main » function in the FreeMarker JAR
>> to load a template and a model (as XML or properties, that is why I
>> showed the link). Then create a separate JAR with the UI to load the
>> template and datamodel from the filesystem or define a datamodel
>> interactively, and save the processed template in a file.
>
> The whole functionality, even if it's only a console app with a main
> method, should be in a separate jar, and certainly even in a separate
> project (like we have freemarker-site, freemarker-docgen, and then we
> could have freemarker-tester or something). Since dependency
> management has become common place, it's considered to be a bad
> practice to overpack your jar-s. In fact, something that we consider
> to do in FM3 is exploding freemarker.jar into submodules (like
> freemarker-servet.jar, etc.). Back then it was more practical to make
> a big monolithic jar, but times are changing.
>
> Also, a separate project (again, still under the umbrella of
> FreeMarker and Apache) means that you can do releases independently of
> the FreeMarker (the engine) project releases.
>
> Also note that "main" methods in a library are seen as security
> problems nowadays. We already have a bunch of them in freemarker.jar,
> and I'm just waiting for the right moment to get rid of them.
>
> So, yeah, I say, separate jar, separate project.
>

-- 
Thanks,
 Daniel Dekany

Reply via email to