Thread overview | |||||||
---|---|---|---|---|---|---|---|
|
May 10, 2011 [dmd-internals] Consistency with MODEL | ||||
---|---|---|---|---|
| ||||
I would point out that both druntime and phobos use 32 and 64 for the values of MODEL for building their 32-bit and 64-bit versions respectively, whereas dmd uses -m32 and -m64. Unless there is a good reason for dmd using -m32 and - m64 instead of 32 and 64, it would probably be best (for consistency's sake, if nothing else), for dmd to use 32 and 64 for MODEL like druntime and phobos do. - Jonathan M Davis |
May 11, 2011 [dmd-internals] Consistency with MODEL | ||||
---|---|---|---|---|
| ||||
Posted in reply to Jonathan M Davis |
On 5/10/2011 7:08 PM, Jonathan M Davis wrote:
> I would point out that both druntime and phobos use 32 and 64 for the values of MODEL for building their 32-bit and 64-bit versions respectively, whereas dmd uses -m32 and -m64. Unless there is a good reason for dmd using -m32 and - m64 instead of 32 and 64, it would probably be best (for consistency's sake, if nothing else), for dmd to use 32 and 64 for MODEL like druntime and phobos do.
>
It's a good idea.
|
May 11, 2011 [dmd-internals] Consistency with MODEL | ||||
---|---|---|---|---|
| ||||
Posted in reply to Walter Bright | On 5/11/2011 7:01 PM, Walter Bright wrote:
>
>
> On 5/10/2011 7:08 PM, Jonathan M Davis wrote:
>> I would point out that both druntime and phobos use 32 and 64 for the values of MODEL for building their 32-bit and 64-bit versions respectively, whereas dmd uses -m32 and -m64. Unless there is a good reason for dmd using -m32 and - m64 instead of 32 and 64, it would probably be best (for consistency's sake, if nothing else), for dmd to use 32 and 64 for MODEL like druntime and phobos do.
>>
>
> It's a good idea.
It's something I've wanted to do for a while. I'll cook up the pull request, but after the next release. It'll disrupt the auto testers so I'd like to have a chance to prepare them for the change so that the period of brokenness can be minimized.
Later,
Brad
|
May 11, 2011 [dmd-internals] Consistency with MODEL | ||||
---|---|---|---|---|
| ||||
Posted in reply to Brad Roberts | On 2011-05-11 21:12, Brad Roberts wrote:
> On 5/11/2011 7:01 PM, Walter Bright wrote:
> > On 5/10/2011 7:08 PM, Jonathan M Davis wrote:
> >> I would point out that both druntime and phobos use 32 and 64 for the values of MODEL for building their 32-bit and 64-bit versions respectively, whereas dmd uses -m32 and -m64. Unless there is a good reason for dmd using -m32 and - m64 instead of 32 and 64, it would probably be best (for consistency's sake, if nothing else), for dmd to use 32 and 64 for MODEL like druntime and phobos do.
> >
> > It's a good idea.
>
> It's something I've wanted to do for a while. I'll cook up the pull request, but after the next release. It'll disrupt the auto testers so I'd like to have a chance to prepare them for the change so that the period of brokenness can be minimized.
It would certainly be best to do it in a manner that avoids causing major problems, so I agree that it makes sense to wait, but we definitely should try and make our makefiles more consistent across the three projects where appropriate. druntime also needs to have a couple of things changed to match up with Phobos (at minimum, it should be putting 32-bit and 64-bit builds in separate directories like Phobos does; and it would also be good with it had an OS=win32wine build like Phobos does). Overall, what we have works fairly well, but there are definitely a few places which where we could greater consistency than we do. In any case, waiting until after the release makes good sense.
- Jonathan M Davis
|
May 11, 2011 [dmd-internals] Consistency with MODEL | ||||
---|---|---|---|---|
| ||||
Posted in reply to Brad Roberts |
On 5/11/2011 9:12 PM, Brad Roberts wrote:
>
> It's something I've wanted to do for a while. I'll cook up the pull request, but after the next release. It'll disrupt the auto testers so I'd like to have a chance to prepare them for the change so that the period of brokenness can be minimized.
>
I agree. Not for this release.
|
Copyright © 1999-2021 by the D Language Foundation