Contributing to the compiler? Tracking suggestions?

[Brad Roberts]

Bill Baxter wrote:
> I take it there's not enough DMD code provided to actually compile 
> anything, right? There at least aren't any makefiles...  GDC is 
> available for hacking, but how useful is a patch against GDC to Walter?
> 
> In general, what are the best ways to help keep this D-train on chugging 
> forward?  Are compiler patches useful?  Library code?  Apps?  Wrappers? 
>  Bug reports?
> 
> I guess what I'm looking for is a big fat "How can I help?" page 
> somewhere.  It would make a good top-level page for Wiki4D, for instance.
> 
> As far as suggestions go, what's the best way to keep them from getting 
> lost?  Wiki4D has a "proposals" section [1], but it's not very populated 
> -- only 5, one of which I added, which is quite few considering that 
> about that many suggestions show up on the newsgroup every week.
> 
> There's also the tracker for D [2], but a comment on one enhancement 
> proposal I saw there says that the D issue tracker is reserved for 
> suggestions that received positive feedback from Walter, but many good 
> ideas are just ignored by Walter until like 90% the community jumps up 
> and down about it simultaneously.
> 
> Finally there's the Unofficial D Wish List [3], but that seems to be 
> rather out of date/seldom pruned, as it includes old things like "add 
> opIn", duplicates ("System.Windows.Forms"), and just irrelevant things 
> like "This list makes a difference?".
> 
> --bb
> 
> [1]  http://www.prowiki.org/wiki4d/wiki.cgi?IdeaDiscussion
> [2]  http://d.puremagic.com/issues/
> [3]  http://www.all-technology.com/eigenpolls/dwishlist/

You'll probably get a bunch of responses because everyone has their own 
opinion of "that one missing thing that makes D not yet useful enough" 
for them.  So, the best answer is probably to work on the things that 
make it useful for you.. or better put, just use it. :)

If you really want a list, there's a couple other pages in the wiki4d 
that are worth a look:

     http://www.prowiki.org/wiki4d/wiki.cgi?HelpDProgress
     http://www.prowiki.org/wiki4d/wiki.cgi?StalePages

Probably the biggest things, in my opinion, are stale or incomplete 
docs. Shortage of introductory walk-through materials for various 
aspects of the language and compiler targeted at fledgling programmers. 
  There's a lot of language comparisons and reference material, but 
those tend to presume pre-existing knowledge.

Restated.. polish is one of the areas that is subtle but important and 
an area that could use some focus.  Pick either a portion of 
www.digitalmars.com/d or wiki4d and give it a thorough review from a 
number of angles:

     - accuracy
         - does it still match the language as implemented
     - completeness
         - what's missing
         - references linked to where relevant
     - clarity
         - how much context is required to comprehend the info
         - is it presented well
         - is it grammatically correct and spelled correctly
     ... almost certainly others

To answer some of your direct questions.  Patches that are contained 
strictly within the front-end files of gdc are useful.  That part of the 
code is shared between dmd and gdc.  Bug fixes more useful than new 
features, imho, but experimentation with new features has it's place too.

Strictly a personal opinion: it's not the language or the compiler that 
needs the most attention.. it's critical mass of people just using it. 
Don't use D for the sake of D, use it for your next project and show 
that it's been useful.  The more people that are showing the viability 
and utility of the language in real work the better.  A lot of time is 
being spent by a number of people in a bit of a closed feedback cycle: 
producing bits and pieces of the compiler, runtime library, or other 
libraries.  This grows the size of the D base, certainly, but I'd dearly 
love to see some people breaking out of that cycle and producing visible 
case studies of how an end product was produced on top of that base and 
how D was clearly the right choice of tools to have used.

Later,
Brad