D should disallow forward references

Don nospam at nospam.com
Wed Aug 26 00:37:23 PDT 2009


bearophile wrote:
> Disallowing forward references, just because it doesn't work well yet, is bad. Most D features are buggy, but I don't want them removed.
> 
> Don:
>> The real problem is that there's about a thousand bugs in bugzilla!!<
> 
> A partial solution is to have more people on removing bugs.
> For example LDC developers seems able, willing and happy to remove some bugs.
> 
> For example some of the bugs absent in LDC:
> 
> LDC fixes some bugs that still exist in DMD. This can make code that works with DMD fail with LDC and vice versa.
>     * 313, 314: import visibility
>     * 1429: == compares associative array equality
>     * 2206: unnamed template mixin of class inside function or class has incorrect classinfo and mangleof
>     * 2716: auto should never mean scope
>     * 2894: abstract classes sometimes allow non-abstract bodyless functions
>     * 2960: CTFE rejects static array to dynamic array casts
> 
> Similar people have to *encouraged* (much more than it's currently done) to help fix bugs in the common front-end. Problems and resistances that stop similar people to contribute to bug fixes have to be found and then removed or reduced. Walter can improve this situation, that's mostly political, group-related, etc, and not much technical.
> My experience about Open Source group dynamics is limited, but even with my ignorance I can see there are things to be improved on such levels too.

Yes, this is a good point. One of the reasons why Linux was allegedly so 
successful in fostering developers was that it made releases so 
frequently (often daily!). If you can see your patches being 
incorporated, you're far more likely to do contribute again.
Currently there are NINETY bugs in bugzilla with patches. This makes 
'patch' one of the largest categories of bugs!

For the last month or so, there's been a DMD source repository on 
dsource (thanks Brad!). Unfortunately, Walter isn't actually using it 
yet, and the method for incorporating patches does not seem to be very 
efficient.

I have a copy of Walter's internal DMD test suite, so I could actually 
test all of the patches and be 'patchmeister'(I'm currently responsible 
for more than half the patches, anyway).
Maybe we could have a 'patchdmd' branch in the repository, which I would 
have write access to, maybe that would make it easier for Walter to 
incorporate patches (it'd be particularly valuable for DMD1, I think). 
But the last thing I'd want to do is make a fork of DMD.



More information about the Digitalmars-d mailing list