[plt-dev] language dialog

On Fri, Jan 29, 2010 at 1:39 PM, Carl Eastlund <carl.eastlund at gmail.com> wrote:
> On Fri, Jan 29, 2010 at 2:32 PM, Robby Findler
> <robby at eecs.northwestern.edu> wrote:
>> On Fri, Jan 29, 2010 at 1:29 PM, Carl Eastlund <carl.eastlund at gmail.com> wrote:
>>> On Fri, Jan 29, 2010 at 2:22 PM, Robby Findler
>>> <robby at eecs.northwestern.edu> wrote:
>>>> On Fri, Jan 29, 2010 at 1:17 PM, Carl Eastlund <carl.eastlund at gmail.com> wrote:
>>>>> Could we populate the Language dialog with popular choices (PLT
>>>>> Scheme, R6RS Scheme, etc.) that all just dump the user into what we
>>>>> know as the Module language, but with the right first line provided?
>>>>> It seems like what users want is split up differently from how we've
>>>>> implemented it, so why not present a menu split up the way users want
>>>>> and keep the implementation split up the way we find convenient.
>>>>
>>>> I considered this, but am afraid that it may be too big of a change
>>>> for some of our users at this point. Also, I'm afraid that we don't
>>>> yet have (up to snuff) #lang-based versions of all of our languages.
>>>
>>> I'm not proposing removing any of our current options, nor using the
>>> "bait and switch" technique for any language except the ones that we
>>> only support via #lang.  The student languages should stay separate
>>> from Module, and anyone who wants Module should be able to get it (by
>>> that or another descriptive name).  But some languages people are
>>> looking for, that they might not think fall under "Module", I don't
>>> see any reason not to provide by name: PLT Scheme, Typed Scheme, R6RS,
>>> etc.
>>
>> The only language I see in the language dialog that appears to meet
>> your criterion is Lazy Scheme. How about we just remove it from the
>> language dialog instead? Or are there others?
>
> You certainly can, but that is the opposite of my suggestion.  It
> seems like the source of confusion is that people are used to
> different languages being different "modes", rather than different
> source programs in one "multiple-language" grammar.  The new dialog is
> a partial step toward helping them out, by making it clearer that "no,
> really, you can get any language you want with #lang".
>
> But I am suggesting, for those who don't immediately "get it", why not
> provide them with what they expect.  Add options for Typed Scheme, PLT
> Scheme, R6RS, Lazy Scheme, and any other #lang-based languages we can
> think of, that we implementers think of as "part of the Module
> language", but users prefer to think of as "a language level".
> There's no reason to force them to think our way about it.  Is there?

Yes, I understood your suggestion. I should have replied more carefully.

I'm saying that I think something like the language dialog is
ultimately doomed and that we should be doing something else entirely.
Until we figure out what that is, I don't think that it is worth our
time and energy to make the language dialog into a kind of
"#lang"-line wizard.

Robby