[rules-dev] removing DSLs

[Wolfgang Laun]

Providing a generic guided editor for unrestrained DSL definitions with
the option of revamping these definitions while maintaining the DRLs
is probably theoretically impossible, at least based on the DSL-to-DRL
translation process as it is currently implemented. Just consider a
mapping of two different DSL phrases to the same DRL pattern:
E.g. (never mind the syntax):

There is a buggy => Car( type == "Buggy" )
# -- and --
There is a car => Car()
- with type {type} => type == "{type}"

Memorizing the translation steps might solve this problem.
-W

On 29/01/2013, Mark Proctor <mproctor at codehaus.org> wrote:
> Btw the other reasons it comes up is tooling, in guvnor. People want to be
> able to change their DSLs, and have the guided editors updated to reflect
> this, without having to update each DSL manually. And then their is the
> request for nested, scoped and related DSL fragments.
>
> Mark
>
>
> On 29 Jan 2013, at 05:43, Mark Proctor <mproctor at codehaus.org> wrote:
>
>> One of the reasons this came up was due to round tripping from the guided
>> editor to drl text format. This is so the guided editor does not use it's
>> own proprietary xml format, but instead uses the .drl directly.
>>
>> Round tripping DRL is not so hard. Round tripping DSLs is much harder, and
>> I suspect will be hard to get fool proof regex. Also we would at a minimum
>> require an escape in the DRL to delimit a DSL sentence.
>>
>> If you think DSLs are useful, I think we can leave them in, with the
>> caveat of the escape. Sound ok?
>>
>> Mark
>> On 29 Jan 2013, at 05:30, Wolfgang Laun <wolfgang.laun at gmail.com> wrote:
>>
>>> Mark,
>>>
>>> in my talk at RulesFest 2011 I've demonstrated that DSLs in their
>>> current form are
>>> indeed useful, perhaps not quite as easy to use as marketing hype
>>> promises.  The current DSL process exploits regular expressions in a
>>> clever way, but this has limits. (Programming in something close to
>>> natural language has been an ongoing
>>> dream since Rear Admiral G. Hopper's fine achievement, but no
>>> remarkable progress has been made in half a century.)
>>>
>>> I have presented DSL rules more than once to an audience, and it seems
>>> that getting non-geeks to appreciate rules is made easier this way.
>>>
>>> There is nothing in the current DSL that depends on DRL except the
>>> undisputed keywords delimiting a rule. I can see no reason why DSL
>>> shouldn't just stay the way it is: the documentation is fairly complete
>>> (much
>>> more than for some other Expert features) and it hasn't needed many
>>> bugfixes I'm aware of.
>>>
>>> Cheers
>>> Wolfgang
>>>
>>> On 29/01/2013, Mark Proctor <mproctor at codehaus.org> wrote:
>>>> How would people feel if we removed DSLs from 6.0? There is no decision
>>>> either way, but I wanted to see if people liked or disliked the idea.
>>>>
>>>> My reason for this is I don't believe DSLs in their current form,
>>>> beyond
>>>> demo ware, are useful. They need a lot more work to turn them into
>>>> guided
>>>> structured documents, we don't have the people to focus on that right
>>>> now,
>>>> and no one from the community has taken this on.
>>>>
>>>> I'd rather see them removed, until they can be done properly.
>>>>
>>>> Mark
>>>> _______________________________________________
>>>> rules-dev mailing list
>>>> rules-dev at lists.jboss.org
>>>> https://lists.jboss.org/mailman/listinfo/rules-dev
>>>>
>>> _______________________________________________
>>> rules-dev mailing list
>>> rules-dev at lists.jboss.org
>>> https://lists.jboss.org/mailman/listinfo/rules-dev
>>
>
>
> _______________________________________________
> rules-dev mailing list
> rules-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/rules-dev
>