[jbosstools-dev] Re: Richfaces 3.2 ?

[Alexey Kazakov]

Hi Sergey,

There are some changes in TLDs in RichFaces 3.2.2 but the version of it 
is still 3.2.1
So it painful for JBoss Tools team to provide proper support of both 
versions (3.2.1 and 3.2.2) in Code Assist for Facelets.
We are going to refactor our Code Assist and use real TLD but not only 
our special XML so it could help us to handle such mismatches
but now it is a real problem. So we have to choose one TLD and use it 
for all 3.2.* richfaces libs in JBoss Tools 3.0.0CR1.
Max, should we replace 3.2.1 by 3.2.2 in our KB plug-in for 3.0.0CR1?

Sergey Smirnov wrote:
> We have never been change this number inside tld. It was 1.2 from the 
> very first version. Mainly, because it does not make any since for 
> run-time. We store the true version in the manifest.mf located close 
> to tlds files inside the META-INF instead.
> Actually, the standard  limits the content of this tag. It must only 
> numbers divided by up to 3 dots. So, we cannot put the exact version 
> there like 3.2.0.GA or 3.2.0.SP1
>
> So, starting with RichFaces 3.2.1, we will turn CDK generator to 
> generate three number divided by dots. It is not ideal, but close to.
>
> In general, we can enhance CDK to generate not only TLD, but the 
> meta-data for code extended assist. In this way, JBDS just needs to 
> take this meta-file from the jar file instead of the place it takes 
> now. It will help to migrate from version to version more smoothly and 
> without extra work from the JBDS team.
>
> I told with Alexey about this feature, but looks like this topic was 
> just forgotten between the other more actual themes on that moment.
>
> ----- Original Message ----- From: "Max Rydahl Andersen" 
> <max.andersen at redhat.com>
> To: "Alexey Kazakov" <akazakov at exadel.com>
> Cc: <jbosstools-dev at lists.jboss.org>; "Sergey Vasilyev" 
> <svasilyev at exadel.com>; "Sergey Smirnov" <sim at exadel.com>
> Sent: Wednesday, May 07, 2008 10:25 AM
> Subject: Re: Richfaces 3.2 ?
>
>
>>>> How long time would it take to add code completion support for RF 
>>>> 3.2 ?
>>>>
>>> If we want to have RF 3.1.x by default (if we can't recognize the
>>> version of lib) then there will be a problem.
>>
>> But isn't the schemas distinct enough to always recognize the correct 
>> version ?
>>
>> Note: if we can't recognize the version i'm probably fine by falling 
>> back to 3.2 by default.
>> btw. why is hard to set a specific version as the default ? Is it 
>> hardcoded to take the latest version as default or ?
>>
>>> Richaces TLD version tag has not been updated since 1.2.
>>> So we are not able to tell one from the other.
>>
>> Are you telling me the richfaces team does not update their TLD's ?
>> I thought the CDK where supposed to make that "easy" ?
>>
>> I've cc'ed in Sergey S. to get his opinion on how we should go about 
>> supporting
>> updates to richfaces if the libraries does not maintain their schema 
>> version id's..?
>>
>>> It would take about one day to provide code completion for RF 3.2 but
>>> only default lib will work.
>>
>> ?
>>
>> /max 
>