pygccxml-development

[pygccxml-development] DSL challenge

[Roman Y. <rom...@gm...>]

Hi. I created document that describes the problems( challenges )  I
have with adding
DSL to Py++.

http://svn.sourceforge.net/viewvc/pygccxml/pyplusplus_dev/docs/peps/dsl_challenge.rest?view=markup

Obviously I need a help, to solve them. Comments, thought and
suggestions are welcome.

-- 
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/

Re: [pygccxml-development] DSL challenge

[Allen B. <al...@vr...>]

Roman Yakovenko wrote:

>Hi. I created document that describes the problems( challenges )  I
>have with adding
>DSL to Py++.
>
>http://svn.sourceforge.net/viewvc/pygccxml/pyplusplus_dev/docs/peps/dsl_challenge.rest?view=markup
>
>Obviously I need a help, to solve them. Comments, thought and
>suggestions are welcome.
>  
>
My first comment would be to put this on the wiki so we can all comment 
on it and discuss it.

The second comment I would have is it seems strange to me that you 
dismiss the idea of a DSL by introducing a proposed API with 
deficiencies that you don't like and they saying "see it is bad".  I 
think there is a way around each of these issues and from what I see of 
the interface you created I already submitted a ~5 line patch that you 
rejected that would add this ability to the existing interface.

The last comment is that I still don't see how you think the template 
support could be done this way with gccxml and pygccxml in the mix.  I 
don't see a  way around the two phased approach.

Anyway, if you move this to the wiki for discussion I will comment on it 
directly.

-Allen

Re: [pygccxml-development] DSL challenge

[Roman Y. <rom...@gm...>]

On 8/22/06, Allen Bierbaum <al...@vr...> wrote:
> My first comment would be to put this on the wiki so we can all comment
> on it and discuss it.

Please become a leader of this effort :-).

> The second comment I would have is it seems strange to me that you
> dismiss the idea of a DSL by introducing a proposed API with
> deficiencies that you don't like and they saying "see it is bad".

The only API I introduced was Py++ one. I did not say anything about how DSL
API looks like.

>  I
> think there is a way around each of these issues and from what I see of
> the interface you created I already submitted a ~5 line patch that you
> rejected that would add this ability to the existing interface.

Lets talk use cases. I introduced the problematic ones. I expect from you
to prove me being wrong or to provide good solution.

> The last comment is that I still don't see how you think the template
> support could be done this way with gccxml and pygccxml in the mix.  I
> don't see a  way around the two phased approach.

This paper I still have to write :-). But in general you are right, I
don't like two
phased approach, but it is possible to implement it.


-- 
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/

Re: [pygccxml-development] DSL challenge

[Matthias B. <ba...@ir...>]

Allen Bierbaum wrote:
> Roman Yakovenko wrote:
> 
>> Hi. I created document that describes the problems( challenges )  I
>> have with adding
>> DSL to Py++.
>>
>> http://svn.sourceforge.net/viewvc/pygccxml/pyplusplus_dev/docs/peps/dsl_challenge.rest?view=markup
>>
>> Obviously I need a help, to solve them. Comments, thought and
>> suggestions are welcome.
>>  
> My first comment would be to put this on the wiki so we can all comment 
> on it and discuss it.

Erm, I still cannot access the wiki....

> The second comment I would have is it seems strange to me that you 
> dismiss the idea of a DSL by introducing a proposed API with 
> deficiencies that you don't like and they saying "see it is bad".  

That was also my impression.... ;)

- Matthias -

Re: [pygccxml-development] DSL challenge

[Allen B. <al...@vr...>]

Matthias Baas wrote:

>Allen Bierbaum wrote:
>  
>
>>Roman Yakovenko wrote:
>>
>>    
>>
>>>Hi. I created document that describes the problems( challenges )  I
>>>have with adding
>>>DSL to Py++.
>>>
>>>http://svn.sourceforge.net/viewvc/pygccxml/pyplusplus_dev/docs/peps/dsl_challenge.rest?view=markup
>>>
>>>Obviously I need a help, to solve them. Comments, thought and
>>>suggestions are welcome.
>>> 
>>>      
>>>
>>My first comment would be to put this on the wiki so we can all comment 
>>on it and discuss it.
>>    
>>
>
>Erm, I still cannot access the wiki....
>  
>
Fixed.  Let me know if it happens again.  I think you probably got hit 
by editing a page quite a few times within a short window.  The system 
thought you were a bot spamming the wiki. :)

-Allen

>  
>
>>The second comment I would have is it seems strange to me that you 
>>dismiss the idea of a DSL by introducing a proposed API with 
>>deficiencies that you don't like and they saying "see it is bad".  
>>    
>>
>
>That was also my impression.... ;)
>
>- Matthias -
>
>
>-------------------------------------------------------------------------
>Using Tomcat but need to do more? Need to support web services, security?
>Get stuff done quickly with pre-integrated technology to make your job easier
>Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
>http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
>_______________________________________________
>pygccxml-development mailing list
>pyg...@li...
>https://lists.sourceforge.net/lists/listinfo/pygccxml-development
>
>  
>

Re: [pygccxml-development] DSL challenge

[Matthias B. <ba...@ir...>]

Allen Bierbaum wrote:
>> Erm, I still cannot access the wiki....
>>
> Fixed.  Let me know if it happens again.  

It works again, thanks!

> I think you probably got hit 
> by editing a page quite a few times within a short window.  

That can happen. I like to preview the page rather often to see if the 
markup is correct.

- Matthias -

Re: [pygccxml-development] DSL challenge

[Roman Y. <rom...@gm...>]

On 8/22/06, Matthias Baas <ba...@ir...> wrote:
> > The second comment I would have is it seems strange to me that you
> > dismiss the idea of a DSL by introducing a proposed API with
> > deficiencies that you don't like and they saying "see it is bad".
>
> That was also my impression.... ;)

Not every day I can quote myself :-):
"...I don't argue, that the second way is better. ..."

It will take few hours to implement this feature. I see my role in
this feature as a
"bad guy" :-), sorry. I will bring difficult use cases and you will
have to solve them.
Thus we will better understand the problem and implement best solution.

-- 
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/

Re: [pygccxml-development] DSL challenge

[Allen B. <al...@vr...>]

Roman Yakovenko wrote:

> On 8/22/06, Allen Bierbaum <al...@vr...> wrote:
>
>> My first comment would be to put this on the wiki so we can all comment
>> on it and discuss it.
>
>
> Please become a leader of this effort :-).

No problem.  I will take on the DSL design and discussion.  Can I assume 
that means that you aren't going to introduce a new DSL and surprise us 
all?  :)

>
>> The second comment I would have is it seems strange to me that you
>> dismiss the idea of a DSL by introducing a proposed API with
>> deficiencies that you don't like and they saying "see it is bad".
>
>
> The only API I introduced was Py++ one. I did not say anything about 
> how DSL
> API looks like.

The BUI that you proposed was what I was thinking about.

>
>>  I
>> think there is a way around each of these issues and from what I see of
>> the interface you created I already submitted a ~5 line patch that you
>> rejected that would add this ability to the existing interface.
>
>
> Lets talk use cases. I introduced the problematic ones. I expect from you
> to prove me being wrong or to provide good solution.

Yes, I have some ideas here.  I will try them out on my current project 
with the method I suggested of extending and overriding the py++ api.

>> The last comment is that I still don't see how you think the template
>> support could be done this way with gccxml and pygccxml in the mix.  I
>> don't see a  way around the two phased approach.
>
>
> This paper I still have to write :-). But in general you are right, I
> don't like two
> phased approach, but it is possible to implement it.

By "paper" can I assume you mean comment on the wiki page about 
templates?  That would be much more helpful IMHO then a .rest document 
or mailing list posting.

-Allen

>
>

Re: [pygccxml-development] DSL challenge

[Allen B. <al...@vr...>]

Allen Bierbaum wrote:

>Roman Yakovenko wrote:
>
>  
>
>>On 8/22/06, Allen Bierbaum <al...@vr...> wrote:
>>
>>    
>>
>>>My first comment would be to put this on the wiki so we can all comment
>>>on it and discuss it.
>>>      
>>>
>>Please become a leader of this effort :-).
>>    
>>
>
>No problem.  I will take on the DSL design and discussion.  Can I assume 
>that means that you aren't going to introduce a new DSL and surprise us 
>all?  :)
>  
>
The current discussion page is here: 
https://realityforge.vrsource.org/view/PyppApi/DslDiscussion

-Allen

Re: [pygccxml-development] DSL challenge

[Roman Y. <rom...@gm...>]

On 8/22/06, Allen Bierbaum <al...@vr...> wrote:
> No problem.  I will take on the DSL design and discussion.  Can I assume
> that means that you aren't going to introduce a new DSL and surprise us
> all?  :)

Don't you like surprises :-)? No. I am pretty happy with current API.
I am not in
a harry to "fix" the situation. I just hope that we will be able to
solve this issue
and implement it before next release.

> > The only API I introduced was Py++ one. I did not say anything about
> > how DSL
> > API looks like.
>
> The BUI that you proposed was what I was thinking about.

It is not an API, but usage example

> > Lets talk use cases. I introduced the problematic ones. I expect from you
> > to prove me being wrong or to provide good solution.
>
> Yes, I have some ideas here.  I will try them out on my current project
> with the method I suggested of extending and overriding the py++ api.

:-( wrong answer. Please don't write single line of code before we solve the
problematic use cases "on paper". I would like to see definitions, before
any implementation.

> > This paper I still have to write :-). But in general you are right, I
> > don't like two
> > phased approach, but it is possible to implement it.
>
> By "paper" can I assume you mean comment on the wiki page about
> templates?  That would be much more helpful IMHO then a .rest document
> or mailing list posting.

No, I mean  rest document. I will write an initial paper and then we will take
the discussion to the wiki.

-- 
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/

Re: [pygccxml-development] DSL challenge

[Allen B. <al...@vr...>]

Roman Yakovenko wrote:

> On 8/22/06, Allen Bierbaum <al...@vr...> wrote:
>
>> No problem.  I will take on the DSL design and discussion.  Can I assume
>> that means that you aren't going to introduce a new DSL and surprise us
>> all?  :)
>
>
> Don't you like surprises :-)? No. I am pretty happy with current API.
> I am not in
> a harry to "fix" the situation. I just hope that we will be able to
> solve this issue
> and implement it before next release.

Okay.

>
>> > The only API I introduced was Py++ one. I did not say anything about
>> > how DSL
>> > API looks like.
>>
>> The BUI that you proposed was what I was thinking about.
>
>
> It is not an API, but usage example

I guess so.  It just seemed that the "usage example" ended up defining a 
proposed API as part of the usage. :)

>
>> > Lets talk use cases. I introduced the problematic ones. I expect 
>> from you
>> > to prove me being wrong or to provide good solution.
>>
>> Yes, I have some ideas here.  I will try them out on my current project
>> with the method I suggested of extending and overriding the py++ api.
>
>
> :-( wrong answer. Please don't write single line of code before we 
> solve the
> problematic use cases "on paper". I would like to see definitions, before
> any implementation.

Well I think we are too late now.  I have already started implementing 
because I have to get something working ASAP.  I need to get my project 
done that I am working on with pyplusplus.  As much as I may like the 
idea of pyplusplus and contributing to it, right now I need to get work 
done to pay the bills.

Nothing in the API will be fixed or anything but it will let me give a 
concrete example of how some of the issues could be fixed.

That said, at the same time I am interested in talking through the 
issues using the wiki.


>> > This paper I still have to write :-). But in general you are right, I
>> > don't like two
>> > phased approach, but it is possible to implement it.
>>
>> By "paper" can I assume you mean comment on the wiki page about
>> templates?  That would be much more helpful IMHO then a .rest document
>> or mailing list posting.
>
>
> No, I mean  rest document. I will write an initial paper and then we 
> will take
> the discussion to the wiki.

Why write it as a .rest document at all?  That really just makes it 
harder for us to get access to it and comment on the contents.

-Allen