if you see the results already, then the system thinks you have already voted.

OK, I didn't think the VTL was that low. Probably because I have never poked around in the eclipse stuff. And your point is well taken - that it would take bad domain/data models to effectively test.

I had an idea once to create a domain randomizer, something that would just spit out large random instances of Configuration.

I have absolutely no idea why this thing thinks I voted.

reveng != templates.

reveng = the process that reads from jdbc metadata and generate a hibernate configuration.

code generation = templates that takes a hibernate configuraiton and dump stuff

reveng is 15%.
code generation is 50-70% of the code base
the rest is the ui, query console and visualization

more db modles = test of reveng
more hbm/java files = test of code generation

So usage of VTL is *very* high since it is the main part of the code generation.

hmm ... so it is a lot of work. when I read 'reveng', I think jdbcconfiguration in combination *with* a generator of some sort, since this is the only thing I have used HT for at work.

No, reveng is just the reveng part.

You are not *required* to use the jdbcconfiguration to generate code - that is the beauty of it all ;)

And no, it is not a lot of work to port - it is a lot of work to keep spending time on stupid velocity exception handling.

do you have any idea about when your going to do this?

after the next release. why ? ;)

I have some time this weekend. I'm planning on dusting off the ol' patch ;)

ah - well i won't make any freemarker conversion over the weekend.

I would suggest bumping the version number up by 1 for this.

3.1.1 ?

the documentation have always contained the following line:

"Note that we expect significant changes to the structure and organization of the built-in templates in one of the next releases."

but yes, i would not normally do this but it is *beta* and nothing prevents you from using the older versions if you want to. ....but i'm open for arguments against doing this move.

Good arguments (and counter-arguments) I know:

1. "I have tons of templates for the tools written velocity style, dont move!"
Answer: there are a very easy migration path via the velocity to freemarker tool and you can use the older versions until you get around porting your templates.
Counter-answer: i will test that tool and if it does not work i'll consider if i can let both engines be supported - but only if i feel that whatever manual conversion is "overwhelming".

2. "I don't like FreeMarker because of X, Y and Z"
Answer: Noone have reported that here in this thread ;)

3...? you got any ?

/max

I would be very suprised if one of those conversion tools ever worked good.

Maybe an abstraction layer for the templating engine ? again, more work :(

I is already abstracted...at least for most parts ,)

dennis - i would not suggest this if i did not believe that the work could be done in a few days. I started this thread to hear if someone had technical arguments *against* FreeMarker or had the magic to make Velocity userfriendly.

Max - if you feel that its the right thing to do just do it. You are the principle here. Given only three people have made a comment then I would guess most people just want the output generated correctly - how its done does not matter. Yes there has been lots of questions about the templates but I am sure they can use the source as an clear and clean example of freemarker.

Hi David,

Don't worry, I planned on doing it (almost) no matter what this disucssion brought up ....just wanted to get a feel if there were many that had an opinion/insight about it and if i should try and support both...apparently not :)

Thanks for you input!