Just wanted to solicit opinion, from y'all _and_ from Walter.

Basically, I feel like there's a momentum building with progress in the language, which is a good thing. I guess this is in part because I'm more involved, and partly because the looming writing of DPD is adding impetus, and partly because Walter seems to be reacting more quickly and positively to criticism.

However, there's still a tendency for threads about specific issues to meander all over the place, and lose focus.

I was wondering whether now might be the time to tighten things up somewhat. I have no specific idea how, and no desire to burden any single individual (especially me!) with a whole load of work, but I feel like there may be a better way for formal language changes and criticisms (=== 'critiques' to my U.S. friends) than mixing things up on the main ng.

Thoughts?

Cheers

Matthew

P.S. FYI: It's been noted that I've been quite aggressive in the last week or two, almost Kris-like on occasion. I know that Walter has a strong back, so this is not for his benefit, but I'd still like to explain so you don't all think I'm just a grumpy bastard who's carping for the sake of it. Ironically, it reflects the fact that I'm finally getting a good feeling about D, in terms of what it is turning into, its usefulness, and its likely 'release' into the real world. Naturally, in writing DPD, I am very keen not to be have to be describing things that I think are embarassingly stupid or ill-considered, so am probably leaking positive fervour into negative postings.

In article <d0o79t$mu2$1@digitaldaemon.com>, Matthew says...
>
>Thoughts?

In short: we need to establish basic criteria for submitting a language change.

(*short-short version at end)

My understanding of the "language development process" in this group was one of consensus, weighted by experience, in the eyes of Walter himself.  I think that the experiment put forward with adding '$' in 0.116 did a good job of illustrating  how this process works.

If we're *all* looking to get our opinion out, contribute, and do so in an orderly fashion, perhaps we need to raise the bar for language changes/suggestions?

I think we need to start taking suggestions as *suggestions* and give more attention to proposals that have been researched and investigated before they're put forward for "review".

I'm not suggesting that we all suddenly put on our Booch/Jacobson/Rumbaugh hats and crank out whitepapers all afternoon.  I am suggesting that attention be given to only those ideas that have taken the following into account:

- Impact to the lexer/parser
- Impact to the developer's behavior and habits
- Impact to existing code

.. and posted this information *succinctly* to the thread.  And by "Impact" I really mean "pros and cons"; everything, no one-sided content allowed.

The actual *change* comes about from the information provided in the 'proposal'. It automatically frames the discussion to follow it as its already put up a good argument for and against.

The rest comes from good reader habits, and modifying the proposal as it lives through the thread.  If you have something to say about a feature proposal, append your consequences to the text.  If you don't care if Walter sees your thoughts or not, then leave it out of the modified proposal.

*To Sum Up:
- Language changes should be proposed formally to the group, along with all the
design consequences (for and against).
- Said proposals are subject to review, commentary and modification by the DNG.
- Walter is free to do whatever he pleases with these 'living' proposals, and
can simply walk to the end of the thread to see the current 'consensus' on the
proposal.

- EricAnderton at yahoo

"pragma" <pragma_member@pathlink.com> wrote in message news:d0oaok$qf0$1@digitaldaemon.com...
> In article <d0o79t$mu2$1@digitaldaemon.com>, Matthew says...
>>
>>Thoughts?
>
> In short: we need to establish basic criteria for submitting a language change.
>
> (*short-short version at end)
>
> My understanding of the "language development process" in this group was one of consensus, weighted by experience, in the eyes of Walter himself.  I think that the experiment put forward with adding '$' in 0.116 did a good job of illustrating  how this process works.
>
> If we're *all* looking to get our opinion out, contribute, and do so in an orderly fashion, perhaps we need to raise the bar for language changes/suggestions?
>
> I think we need to start taking suggestions as *suggestions* and give more attention to proposals that have been researched and investigated before they're put forward for "review".
>
> I'm not suggesting that we all suddenly put on our Booch/Jacobson/Rumbaugh hats and crank out whitepapers all afternoon.  I am suggesting that attention be given to only those ideas that have taken the following into account:
>
> - Impact to the lexer/parser
> - Impact to the developer's behavior and habits
> - Impact to existing code
>
> .. and posted this information *succinctly* to the thread.  And by "Impact" I really mean "pros and cons"; everything, no one-sided content allowed.
>
> The actual *change* comes about from the information provided in the 'proposal'. It automatically frames the discussion to follow it as its already put up a good argument for and against.
>
> The rest comes from good reader habits, and modifying the proposal as it lives through the thread.  If you have something to say about a feature proposal, append your consequences to the text.  If you don't care if Walter sees your thoughts or not, then leave it out of the modified proposal.
>
> *To Sum Up:
> - Language changes should be proposed formally to the group, along with all the
> design consequences (for and against).
> - Said proposals are subject to review, commentary and modification by the DNG.
> - Walter is free to do whatever he pleases with these 'living' proposals, and
> can simply walk to the end of the thread to see the current 'consensus' on the
> proposal.

All sounds good, apart from the Impact to the lexer/parser. Why should only the subset of the entire potential user community who are well-versed/expert in the ways of compiler technology be the only ones whose proposals are considered?

Matthew wrote:
> Just wanted to solicit opinion, from y'all _and_ from Walter.
> 
> Basically, I feel like there's a momentum building with progress in the language, which is a good thing. I guess this is in part because I'm more involved, and partly because the looming writing of DPD is adding impetus, and partly because Walter seems to be reacting more quickly and positively to criticism.
> 
> However, there's still a tendency for threads about specific issues to meander all over the place, and lose focus.
> 
> I was wondering whether now might be the time to tighten things up somewhat. I have no specific idea how, and no desire to burden any single individual (especially me!) with a whole load of work, but I feel like there may be a better way for formal language changes and criticisms (=== 'critiques' to my U.S. friends) than mixing things up on the main ng.
> 
> Thoughts?
> 
> Cheers
> 
> Matthew
> 
> P.S. FYI: It's been noted that I've been quite aggressive in the last week or two, almost Kris-like on occasion. I know that Walter has a strong back, so this is not for his benefit, but I'd still like to explain so you don't all think I'm just a grumpy bastard who's carping for the sake of it. Ironically, it reflects the fact that I'm finally getting a good feeling about D, in terms of what it is turning into, its usefulness, and its likely 'release' into the real world. Naturally, in writing DPD, I am very keen not to be have to be describing things that I think are embarassingly stupid or ill-considered, so am probably leaking positive fervour into negative postings.
> 
> 
> 


Howzabout a "feature" request type script on the website? Combined with a bug tracker (I believe the commonly are packaged together, right?) And then a voting system for both the features and bugs, showing which should be taken seriously and worked on first, respectively.

-- 
-PIB

--
"C++ also supports the notion of *friends*: cooperative classes that
are permitted to see each other's private parts." - Grady Booch

Ah yes! I repeat: a PEP like system ( http://www.python.org/peps/ ) to track
suggestions and
their arguments!

I'll write the first one, PEP000, being a suggestion to have a pep like system, etc.. OK? But... Is a wiki suitable for this kind-of thing?

L.

Paul Bonser wrote:

> Howzabout a "feature" request type script on the website? Combined with a bug tracker (I believe the commonly are packaged together, right?)

A bugzilla like system to report RFE and Bugs have been suggested...
Repeatedly. But meanwhile, we have the digitalmars.D.bugs newgroup,
and http://www.prowiki.org/wiki4d/wiki.cgi?FeatureRequestList

--anders

"Matthew" <admin.hat@stlsoft.dot.org> wrote in message news:d0ocrp$s66$1@digitaldaemon.com...
>
> "pragma" <pragma_member@pathlink.com> wrote in message news:d0oaok$qf0$1@digitaldaemon.com...
>> In article <d0o79t$mu2$1@digitaldaemon.com>, Matthew says...
>>>
>>>Thoughts?
>>
>> In short: we need to establish basic criteria for submitting a language change.
>>
>> (*short-short version at end)
>>
>> My understanding of the "language development process" in this group was
>> one of
>> consensus, weighted by experience, in the eyes of Walter himself.  I
>> think that
>> the experiment put forward with adding '$' in 0.116 did a good job of
>> illustrating  how this process works.
>>
>> If we're *all* looking to get our opinion out, contribute, and do so in
>> an
>> orderly fashion, perhaps we need to raise the bar for language
>> changes/suggestions?
>>
>> I think we need to start taking suggestions as *suggestions* and give
>> more
>> attention to proposals that have been researched and investigated before
>> they're
>> put forward for "review".
>>
>> I'm not suggesting that we all suddenly put on our
>> Booch/Jacobson/Rumbaugh hats
>> and crank out whitepapers all afternoon.  I am suggesting that attention
>> be
>> given to only those ideas that have taken the following into account:
>>
>> - Impact to the lexer/parser
>> - Impact to the developer's behavior and habits
>> - Impact to existing code
>>
>> .. and posted this information *succinctly* to the thread.  And by
>> "Impact" I
>> really mean "pros and cons"; everything, no one-sided content allowed.
>>
>> The actual *change* comes about from the information provided in the
>> 'proposal'.
>> It automatically frames the discussion to follow it as its already put up
>> a good
>> argument for and against.
>>
>> The rest comes from good reader habits, and modifying the proposal as it
>> lives
>> through the thread.  If you have something to say about a feature
>> proposal,
>> append your consequences to the text.  If you don't care if Walter sees
>> your
>> thoughts or not, then leave it out of the modified proposal.
>>
>> *To Sum Up:
>> - Language changes should be proposed formally to the group, along with
>> all the
>> design consequences (for and against).
>> - Said proposals are subject to review, commentary and modification by
>> the DNG.
>> - Walter is free to do whatever he pleases with these 'living' proposals,
>> and
>> can simply walk to the end of the thread to see the current 'consensus'
>> on the
>> proposal.
>
> All sounds good, apart from the Impact to the lexer/parser. Why should only the subset of the entire potential user community who are well-versed/expert in the ways of compiler technology be the only ones whose proposals are considered?
>

I don't think Eric meant in in that way.. I don't think he was implying that compiler developers themselves should have an inordinate amount of sway in the design of the language, but that the /impact/ on the compiler should be an important consideration.

IMHO, the compiler technology aspect is an important consideration to getting the language ported (relatively quickly) to other platforms and/or to get other vendors at least interested in rolling their own D compiler.

So I think it is valid that those considerations should have at least some reasonable amount of weight in the decision, especially for v1.0.

If something is a kludge to implement in the compiler, it will probably end-up being a kludge for the developer as well. For example: 'vector<vector<...> >' (where you have to have a space in-between ">" & ">").

- Dave

In article <d0ptf3$2h16$1@digitaldaemon.com>, Dave says...
>
>"Matthew" <admin.hat@stlsoft.dot.org> wrote in message news:d0ocrp$s66$1@digitaldaemon.com...
>>
>> "pragma" <pragma_member@pathlink.com> wrote in message
>>> *To Sum Up:
>>> - Language changes should be proposed formally to the group, along with
>>> all the
>>> design consequences (for and against).
>>> - Said proposals are subject to review, commentary and modification by
>>> the DNG.
>>> - Walter is free to do whatever he pleases with these 'living' proposals,
>>> and
>>> can simply walk to the end of the thread to see the current 'consensus'
>>> on the
>>> proposal.
>>
>> All sounds good, apart from the Impact to the lexer/parser. Why should only the subset of the entire potential user community who are well-versed/expert in the ways of compiler technology be the only ones whose proposals are considered?
>>
>
>I don't think Eric meant in in that way.. I don't think he was implying that compiler developers themselves should have an inordinate amount of sway in the design of the language, but that the /impact/ on the compiler should be an important consideration.
>
>IMHO, the compiler technology aspect is an important consideration to getting the language ported (relatively quickly) to other platforms and/or to get other vendors at least interested in rolling their own D compiler.
>
>So I think it is valid that those considerations should have at least some reasonable amount of weight in the decision, especially for v1.0.
>
>If something is a kludge to implement in the compiler, it will probably end-up being a kludge for the developer as well. For example: 'vector<vector<...> >' (where you have to have a space in-between ">" & ">").

Dave, thank you for making my point. That's exactly what I was trying to convery.  :)

If you look back at the history of changes made to D, the lex/parse aspects of a language change has everything to do with its parseabilty which in turn effects different aspects of the D toolchain (doxygen for instance).  But you can't design a language change without taking even the most rudimentary parser impacts into account (look at the '**' thread for a solid example of this).

Yes, it does raise the bar for such feature proposals since its not everyone's area of expertise.  However, it shouldn't keep anyone from championing a particular improvment to D even if they have to ask for a little help here and there.

I think there's room enough here for someone to say "here's my idea, but what about the lexer?" and folks will say this or that on the subject.  If it doesn't interest the group, it doesn't ever mature into a feasable proposal.  That's what group-development and peer-review is all about.

*I'll also add that some feature requests aren't really language changes, so parser/lexer impact would be a moot point at best (ie. reflection).


Overall, the notion is to focus our activities from group discussion, to group *development* of ideas.  Ultimately, this is so Walter can merely pluck the low-hanging fruit as these ideas ripen into well-developed concepts.

I honestly don't care exactly *how* its done, but I do feel that its essential to making things work around here.  I also think that my original post is a good starting point.


Provided that something like this is the way forward: should we wait to have a PEP-like system in place, or should we go ahead and start using the newsgroup here for the review process instead?

- EricAnderton at yahoo

I think Walter needs to lay down the official language rules for 1.0, 1.1, and 2.0.  Some may not like them and drop off, but it would be nice if 1.0 were no longer a moving target.  2.0 will be open for discussion, proofs, hissy-fits, etc, until it gets closer.

Once the target stops moving, people, happy or not, can focus on writing books, improved libraries, ports, IDEs, whatever with much more confidence. That means no view towards 2.0 while writing these things.

"Matthew" <admin@stlsoft.dot.dot.dot.dot.org> wrote in message news:d0o79t$mu2$1@digitaldaemon.com...
> Just wanted to solicit opinion, from y'all _and_ from Walter.
>
> Basically, I feel like there's a momentum building with progress in the language, which is a good thing. I guess this is in part because I'm more involved, and partly because the looming writing of DPD is adding impetus, and partly because Walter seems to be reacting more quickly and positively to criticism.
>
> However, there's still a tendency for threads about specific issues to meander all over the place, and lose focus.
>
> I was wondering whether now might be the time to tighten things up somewhat. I have no specific idea how, and no desire to burden any single individual (especially me!) with a whole load of work, but I feel like there may be a better way for formal language changes and criticisms (=== 'critiques' to my U.S. friends) than mixing things up on the main ng.
>
> Thoughts?
>
> Cheers
>
> Matthew
>
> P.S. FYI: It's been noted that I've been quite aggressive in the last week or two, almost Kris-like on occasion. I know that Walter has a strong back, so this is not for his benefit, but I'd still like to explain so you don't all think I'm just a grumpy bastard who's carping for the sake of it. Ironically, it reflects the fact that I'm finally getting a good feeling about D, in terms of what it is turning into, its usefulness, and its likely 'release' into the real world. Naturally, in writing DPD, I am very keen not to be have to be describing things that I think are embarassingly stupid or ill-considered, so am probably leaking positive fervour into negative postings.
>
>
> 

But do you take Kris' point made a few days ago, that in some sense this language now belongs to us all?

I think it's fair to say that many are still some way from a point at which they could accept D 1.0. Now of course Walter can do exactly as he pleases, but I would question the wisdom of releasing 1.0, and thereby laying down in stone things that would subsequently be difficult/impossible to change, at a time when many of the 'D people' consider it unready. It's not a question of blackmail/undue influence/whatever, but just common sense.

For my part, D currently contains sufficient flaws that I would be very unhappy with a 1.0 line in the sand anytime in the immediate future. Pure guestimation inclines me to expect 1.0 sometime around (N.H.) autumn.

The opposing force, if you like, is the need to get it to the point while the momentum's still on the ascendant, and avoid the "it's just a research language" status that it would inevitably receive if it just went on and on and on.

The challenge is in working out the sweet spot.



"Charlie Patterson" <charliep1@excite.com> wrote in message news:d0q958$2u1c$1@digitaldaemon.com...
>I think Walter needs to lay down the official language rules for 1.0, 1.1, and 2.0.  Some may not like them and drop off, but it would be nice if 1.0 were no longer a moving target.  2.0 will be open for discussion, proofs, hissy-fits, etc, until it gets closer.
>
> Once the target stops moving, people, happy or not, can focus on writing books, improved libraries, ports, IDEs, whatever with much more confidence. That means no view towards 2.0 while writing these things.
>
> "Matthew" <admin@stlsoft.dot.dot.dot.dot.org> wrote in message news:d0o79t$mu2$1@digitaldaemon.com...
>> Just wanted to solicit opinion, from y'all _and_ from Walter.
>>
>> Basically, I feel like there's a momentum building with progress in the language, which is a good thing. I guess this is in part because I'm more involved, and partly because the looming writing of DPD is adding impetus, and partly because Walter seems to be reacting more quickly and positively to criticism.
>>
>> However, there's still a tendency for threads about specific issues to meander all over the place, and lose focus.
>>
>> I was wondering whether now might be the time to tighten things up somewhat. I have no specific idea how, and no desire to burden any single individual (especially me!) with a whole load of work, but I feel like there may be a better way for formal language changes and criticisms (=== 'critiques' to my U.S. friends) than mixing things up on the main ng.
>>
>> Thoughts?
>>
>> Cheers
>>
>> Matthew
>>
>> P.S. FYI: It's been noted that I've been quite aggressive in the last week or two, almost Kris-like on occasion. I know that Walter has a strong back, so this is not for his benefit, but I'd still like to explain so you don't all think I'm just a grumpy bastard who's carping for the sake of it. Ironically, it reflects the fact that I'm finally getting a good feeling about D, in terms of what it is turning into, its usefulness, and its likely 'release' into the real world. Naturally, in writing DPD, I am very keen not to be have to be describing things that I think are embarassingly stupid or ill-considered, so am probably leaking positive fervour into negative postings.
>>
>>
>>
>
>