www.digitalmars.com         C & C++   DMDScript  

D.gnu - D.2.062 merge

reply "Iain Buclaw" <ibuclaw ubuntu.com> writes:
I do believe I saw a new release mentioned in announcements. So 
this will be a priority to get done first and finished as soon as 
possible.

Regards
Iain.
Feb 18 2013
parent reply Iain Buclaw <ibuclaw ubuntu.com> writes:
On 18 February 2013 08:16, Iain Buclaw <ibuclaw ubuntu.com> wrote:

 I do believe I saw a new release mentioned in announcements. So this will
 be a priority to get done first and finished as soon as possible.

 Regards
 Iain.

Once again, this is proving to be an [expletive] pain. I know Walter has been tying to help out with refactoring code, but it touches very sensitive areas that break easily. I have it in the works to replace these areas in gdc (see thread on removing toobj, typinf, and todt) - but it looks like things will be worse before they start improving again on this instance. -- Iain Buclaw *(p < e ? p++ : p) = (c & 0x0f) + '0';
Feb 19 2013
parent reply "eles" <eles eles.com> writes:
On Tuesday, 19 February 2013 at 11:33:59 UTC, Iain Buclaw wrote:
 On 18 February 2013 08:16, Iain Buclaw <ibuclaw ubuntu.com> 
 wrote:
   I have it in the works to replace these areas in
 gdc (see thread on removing toobj, typinf, and todt) - but it 
 looks like
 things will be worse before they start improving again on this 
 instance.

Won't be better to merge rather smaller increments? For example, merging one commit (from dmd) at a time? That way, the gdc will follow the dmd commit by commit, and not only release by release. I hope this would bring easier merging, as less code changes are required. What do you think?
Feb 20 2013
next sibling parent Iain Buclaw <ibuclaw ubuntu.com> writes:
On 20 February 2013 08:05, eles <eles eles.com> wrote:

 On Tuesday, 19 February 2013 at 11:33:59 UTC, Iain Buclaw wrote:

 On 18 February 2013 08:16, Iain Buclaw <ibuclaw ubuntu.com> wrote:
   I have it in the works to replace these areas in
 gdc (see thread on removing toobj, typinf, and todt) - but it looks like
 things will be worse before they start improving again on this instance.

Won't be better to merge rather smaller increments? For example, merging one commit (from dmd) at a time? That way, the gdc will follow the dmd commit by commit, and not only release by release. I hope this would bring easier merging, as less code changes are required. What do you think?

That's mid to long term goal (we currently do the same with gcc development, for instance). But not until have dropped the remainder dmd backend dependencies from the frontend. Regards -- Iain Buclaw *(p < e ? p++ : p) = (c & 0x0f) + '0';
Feb 20 2013
prev sibling next sibling parent Iain Buclaw <ibuclaw ubuntu.com> writes:
On 20 February 2013 11:24, Iain Buclaw <ibuclaw ubuntu.com> wrote:

 On 20 February 2013 08:05, eles <eles eles.com> wrote:

 On Tuesday, 19 February 2013 at 11:33:59 UTC, Iain Buclaw wrote:

 On 18 February 2013 08:16, Iain Buclaw <ibuclaw ubuntu.com> wrote:
   I have it in the works to replace these areas in
 gdc (see thread on removing toobj, typinf, and todt) - but it looks like
 things will be worse before they start improving again on this instance.

Won't be better to merge rather smaller increments? For example, merging one commit (from dmd) at a time? That way, the gdc will follow the dmd commit by commit, and not only release by release. I hope this would bring easier merging, as less code changes are required. What do you think?

That's mid to long term goal (we currently do the same with gcc development, for instance). But not until have dropped the remainder dmd backend dependencies from the frontend.

OK, D Frontend is merged. Phobos and Druntime are yet to be updated, so subsequently it doesn't yet build the library yet. TODO: - Add compiler support for new cent/ucent types. - Test CTFE feature for doing reinterpret casts between int <-> float. Should work, but untested. - Phobos / D runtime merge, obviously... - Once above done, blast through re-implementation of toDt, toObjFile routines that removes all dependencies on DMD backend. - Breathe and sigh. Regards -- Iain Buclaw *(p < e ? p++ : p) = (c & 0x0f) + '0';
Feb 20 2013
prev sibling parent Iain Buclaw <ibuclaw ubuntu.com> writes:
On 20 February 2013 18:27, Iain Buclaw <ibuclaw ubuntu.com> wrote:

 On 20 February 2013 11:24, Iain Buclaw <ibuclaw ubuntu.com> wrote:

 On 20 February 2013 08:05, eles <eles eles.com> wrote:

 On Tuesday, 19 February 2013 at 11:33:59 UTC, Iain Buclaw wrote:

 On 18 February 2013 08:16, Iain Buclaw <ibuclaw ubuntu.com> wrote:
   I have it in the works to replace these areas in
 gdc (see thread on removing toobj, typinf, and todt) - but it looks like
 things will be worse before they start improving again on this instance.

Won't be better to merge rather smaller increments? For example, merging one commit (from dmd) at a time? That way, the gdc will follow the dmd commit by commit, and not only release by release. I hope this would bring easier merging, as less code changes are required. What do you think?

That's mid to long term goal (we currently do the same with gcc development, for instance). But not until have dropped the remainder dmd backend dependencies from the frontend.

OK, D Frontend is merged. Phobos and Druntime are yet to be updated, so subsequently it doesn't yet build the library yet. TODO: - Add compiler support for new cent/ucent types. - Test CTFE feature for doing reinterpret casts between int <-> float. Should work, but untested. - Phobos / D runtime merge, obviously... - Once above done, blast through re-implementation of toDt, toObjFile routines that removes all dependencies on DMD backend. - Breathe and sigh.

Phobos/Druntime merged. TODO: - cent/ucent - Test CTFE - re-implementation of toDt, toObjFile. -- Iain Buclaw *(p < e ? p++ : p) = (c & 0x0f) + '0';
Feb 21 2013