On 09/05/12 19:24, David Nadlinger wrote:
This would be great, but at least for LDC, the biggest problem at the moment in that regard is manpower –currently, most of us primarily work on it whenever it doesn't compile our own projects (and when specific bug reports come in, obviously). This works reasonably well, e.g. Alexey merged the 2.059 frontend more than 2 1/2 weeks ago, which is not terribly late (where is GDC right now, btw?), but could quite clearly be improved. At least personally, though, I'd currently find it hard to commit to releasing simultaneously with DMD, because it might entail doing larger amounts of merging/testing work on short notice as long as there isn't at least some kind of semi-formal release schedule for DMD.
I shall have to have a go at building LDC from source, I've been wanting to try out ldc2 for ages.
TBH I was not thinking so much of the LDC team having to do extra work, as the core DMD team doing more to ensure that the frontend updates work across all the different backends. Tricky in the short term given the volume of work that still has to be done, possibly manageable in the longer term as the frontend stabilizes and the number of contributors increases.
The reason for proposing this is that currently if I wish to hack on Druntime or Phobos, I _have_ to use DMD. True parity of the open-source compilers would be contributors being able to use their compiler of choice. That should put all the "not OS" complaints to bed properly.
