View mode: basic / threaded / horizontal-split · Log in · Help
November 30, 2007
CONSTipation
The NG is really getting blocked up with const ideas at the moment... it 
seems to me we need to actually settle down and try the new system 
before making judgements and trying to 'fix' it.

I for one can't really tell yet how the changes will affect the way I 
code my D, so it surprises me that we already have so many opinions on 
how good or bad it is.

Regan
November 30, 2007
Re: CONSTipation
"Regan Heath" <regan@netmail.co.nz> wrote in message 
news:fipi6o$98c$1@digitalmars.com...
> The NG is really getting blocked up with const ideas at the moment... it 
> seems to me we need to actually settle down and try the new system before 
> making judgements and trying to 'fix' it.
>
> I for one can't really tell yet how the changes will affect the way I code 
> my D, so it surprises me that we already have so many opinions on how good 
> or bad it is.
>
> Regan

There's no harm in talking about better ways to do it.  If a proposed idea 
isn't sound, then by all means shoot it down.  But there are certain aspects 
of the syntax that are obvious to me to be less than ideal, and I don't have 
to write a million lines of D code to see that.
November 30, 2007
Re: CONSTipation
On 11/30/07, Regan Heath <regan@netmail.co.nz> wrote:
> The NG is really getting blocked up with const ideas at the moment... it
> seems to me we need to actually settle down and try the new system
> before making judgements and trying to 'fix' it.

I've been using D2.008 since the day it appeared. All of my comments
are based on playing with it and using it, and trying to get the hang
of how it behaves.


> I for one can't really tell yet how the changes will affect the way I
> code my D, so it surprises me that we already have so many opinions on
> how good or bad it is.

That's certainly true. But when you find that

   invariant(int) x = 0;
   x = 1;

compiles and runs, it is surely reasonable to question whether there
might be a better syntax for what are clearly /superb/ concepts.
November 30, 2007
Re: CONSTipation
Regan Heath wrote:

> The NG is really getting blocked up with const ideas at the moment... it
> seems to me we need to actually settle down and try the new system
> before making judgements and trying to 'fix' it.
> 
> I for one can't really tell yet how the changes will affect the way I
> code my D, so it surprises me that we already have so many opinions on
> how good or bad it is.
> 
> Regan

Half of my posts have a comment to the effect of "please document this
better to avoid future confusion on this topic".

Additionally, Walter has said before that people should not just say "this
sucks" and instead should put on the designers hat and tell him how to fix
it.  That, of course, then opens up all of us to criticism and you get
these long threads about how to redesign the language.  When there's no
concensus, we all accept that Walter's solution is good enough.

I've tried to adopt that methodology because I believe it leads to a natural
process that helps D mature as a language.  It may be rare that we convince
Walter (or a large fraction of the D community) that our ideas are correct,
but that doesn't mean it has no value.  

I honestly think that the reception to this latest const has been much
warmer than the last time.  IIRC, there's only been two issues raised.  One
was how ambiguity over how const functions are specified, and the other (by
me) was ambiguity over const and invariant storage classes relative to
const and invariant types.

I really do try to read every other post on the NG prior to making my own
posts, and when I post I try to think about what I'm saying so I don't just
spew out my random thoughts at the time.  It took me a very long time to
draft my "invariant doesn't apply to declared symbols" post.  The content
changed many times.  I honestly consider it a point of confusion...  where
people could use stuff thinking one thing and then be surprised to realize
it means something else.
Top | Discussion index | About this forum | D home