light-table: live programming for python

impressive how much support they got:

http://www.kickstarter.com/projects/ibdknox/light-table       

not much missing in our system, is it?

On 1 June 2012 13:29, Camillo Bruni <[hidden email]> wrote:
> impressive how much support they got:
>
> http://www.kickstarter.com/projects/ibdknox/light-table
>
> not much missing in our system, is it?

No, definitely not. Most of it is automatically pulling up contextual
info (like if you're implementing #myFoo on some object, bring up all
the implementors in some no-clicks-needed window (possibly with some
classes preferentially highlighted - sub/superclasses and other
classes that satisfy similar protocols) and the senders of same in
another (with suitable highlighting of the actual send site). The more
clever de-abstraction stuff we can get quite easily by manipulating
ASTs. (In a Smalltalk-based system, I'd half expect us to keep ASTs
attached to method on which the programmer worked.)

frank

We're not missing much, and somehow Smalltalk and particularly Pharo provides much more.

It's funny how a non-mainstream lang like Clojure was able to pitch a tool like this to raise $300K in Kickstarter. I envy these kind of outcomes, and always makes me thing what we, Smalltalkers as a community, are missing something regarding communication/marketing.

Best regards,

--
Esteban

On Jun 1, 2012, at 5:03 PM, Esteban A. Maringolo wrote:

> We're not missing much, and somehow Smalltalk and particularly Pharo provides
> much more.
>
> It's funny how a non-mainstream lang like Clojure was able to pitch a tool
> like this to raise $300K in Kickstarter. I envy these kind of outcomes, and
> always makes me thing what we, Smalltalkers as a community, are missing
> something regarding communication/marketing.

Yes.

        Marcus


--
Marcus Denker -- http://marcusdenker.de

we should produce more 5 min videos!!!
Look we are not able to even do that.

Stef

On Jun 1, 2012, at 5:02 PM, Esteban A. Maringolo wrote:

Stéphane Ducasse wrote

we should produce more 5 min videos!!!
Look we are not able to even do that.

No smalltalker with a sane mind would record himself five minutes doing his regular work. :)

But truth is we should find a way to impress people like light table did.

The most impressive thing in the demo, and I'd bet that was the main resonator in the audience, was the variables replacement by its current values.

I don't know how hard would be to have debugger with a splitted window or viewing mode showing the code with the actual values of the variables.

Regards,

--
Esteban.

we should get an infrastructure so that it is easy to experiment with such things.
This is why we need OPAL and AST right in the system.

It is time to rethink our system before it gets just too oldish.

Stef

On Jun 1, 2012, at 8:12 PM, Esteban A. Maringolo wrote:

On 1 June 2012 20:12, Esteban A. Maringolo <[hidden email]> wrote: Haha.. well said!


> But truth is we should find a way to impress people like light table did.
>
The truth is, that its not impressed me.. after things i seen in smalltalk.

> The most impressive thing in the demo, and I'd bet that was the main
> resonator in the audience, was the variables replacement by its current
> values.
>
> I don't know how hard would be to have debugger with a splitted window or
> viewing mode showing the code with the actual values of the variables.
>

we should have better text/parser model..
then you can easily view the variable value(s), just by hovering mouse
over the text piece
in debugger.

btw, Gaucho already takes steps into that direction: it using parser
to navigate the code
using the AST tree (so you not walking over separate characters, but
over AST nodes)
it is not far from that point to debugger and/or code pane, where you
can exploit that (code highlighting/contextual help, navigating etc
etc).

--
Best regards,
Igor Stasenko.

Igor Stasenko wrote

On 1 June 2012 20:12, Esteban A. Maringolo <[hidden email]> wrote:
>> But truth is we should find a way to impress people like light table did.
>The truth is, that its not impressed me.. after things i seen in smalltalk.

Not me either, I even ranted about it as soon as it was published (http://goo.gl/7Eaxa). Apparently I have no community impact :P

Igor Stasenko wrote

>> I don't know how hard would be to have debugger with a splitted window or
>> viewing mode showing the code with the actual values of the variables.

>we should have better text/parser model..
>then you can easily view the variable value(s), just by hovering mouse
>over the text piece in debugger.

>btw, Gaucho already takes steps into that direction: it using parser
>to navigate the code
>using the AST tree (so you not walking over separate characters, but
>over AST nodes)
>it is not far from that point to debugger and/or code pane, where you
>can exploit that (code highlighting/contextual help, navigating etc
>etc).

We take a similar approach in our software app, where users program formulas in a pseudo language for which we use SmaCC and hence we have a parse tree.

Then users can see both their "code" with the variables replaced by their values (evaluated with a particular domain context), and even view and browse the expression tree (no one uses it, it just was simple to build). Though it is not in real time because could involve heavy DB queries, it is completely feasible to do with "regular" smalltalk code.

Regards!