Date: Wed, 4 May 2022 13:45:11 +0200
Am 04.05.2022 um 13:33 schrieb Gabriel Dos Reis:
> Where is "implementation partition" defined in the standards text?
>
> -- Gaby, not defending an tool-specific terminology, but genuinely curious
I may have been fallen victim of a mental implicit conversion, but when
parsing the standards text in elevator mode in recent years I came to
the conclusion this particular module TU type is called "module
implementation partition". If this is just a mirage then let's find a
proper name for these beasts or stick with whatever already exists. I
told my audience all the time that these are the modules equivalent to
library-internal headers but that's just one use case.
> -----Original Message-----
> From: SG15 <sg15-bounces_at_[hidden]> On Behalf Of Daniela Engert via SG15
> Sent: Wednesday, May 4, 2022 4:27 AM
> To: sg15_at_[hidden]
> Cc: dani <dani_at_[hidden]>
> Subject: Re: [SG15] P2581R0: Specifying the Interoperability of Binary Module Interface Files
>
> Am 04.05.2022 um 12:46 schrieb Nathan Sidwell via SG15:
>> On 5/3/22 10:32, Ben Boeckel via SG15 wrote:
>>> On Tue, May 03, 2022 at 12:48:24 +0000, Gabriel Dos Reis wrote:
>>>> I have a lot to say in this thread, but a request for the CMake folks:
>>>>
>>>>> CXX_MODULE_HEADERS
>>>> Please, can we avoid "module headers" or "header modules" as
>>>> terminology in the community?
>>>> We've settled on "header units". If we need something more than
>>>> that, please let's work on it.
>>> There are currently three names used:
>>> - `CXX_MODULE_INTERNAL_PARTITIONS`: for any TU which is a partition
>>> without `export` and not an implementation unit (needs to be
>>> separate because of the `-internalPartition` flag is needed during
>>> scanning; it wouldn't be that important if the non-standard MSVC
>>> extention needed the flag and this one did not)
>> IIUC these are what the std calls implementation partitions. It is
>> unfortunate the same name is not being used.
> I'd also prefer this name as coined by the standard instead of a
> tool-specific one.
>
>
> _______________________________________________
> SG15 mailing list
> SG15_at_[hidden]
> https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.isocpp.org%2Fmailman%2Flistinfo.cgi%2Fsg15&data=05%7C01%7Cgdr%40microsoft.com%7Cbc42e15e539f4b71ecbb08da2dc0f3a4%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C637872603993970460%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=PP6%2FjkGQIKfh1FJPFy47qlu8m04El5vgEO8Z7dTz5Uo%3D&reserved=0
> Where is "implementation partition" defined in the standards text?
>
> -- Gaby, not defending an tool-specific terminology, but genuinely curious
I may have been fallen victim of a mental implicit conversion, but when
parsing the standards text in elevator mode in recent years I came to
the conclusion this particular module TU type is called "module
implementation partition". If this is just a mirage then let's find a
proper name for these beasts or stick with whatever already exists. I
told my audience all the time that these are the modules equivalent to
library-internal headers but that's just one use case.
> -----Original Message-----
> From: SG15 <sg15-bounces_at_[hidden]> On Behalf Of Daniela Engert via SG15
> Sent: Wednesday, May 4, 2022 4:27 AM
> To: sg15_at_[hidden]
> Cc: dani <dani_at_[hidden]>
> Subject: Re: [SG15] P2581R0: Specifying the Interoperability of Binary Module Interface Files
>
> Am 04.05.2022 um 12:46 schrieb Nathan Sidwell via SG15:
>> On 5/3/22 10:32, Ben Boeckel via SG15 wrote:
>>> On Tue, May 03, 2022 at 12:48:24 +0000, Gabriel Dos Reis wrote:
>>>> I have a lot to say in this thread, but a request for the CMake folks:
>>>>
>>>>> CXX_MODULE_HEADERS
>>>> Please, can we avoid "module headers" or "header modules" as
>>>> terminology in the community?
>>>> We've settled on "header units". If we need something more than
>>>> that, please let's work on it.
>>> There are currently three names used:
>>> - `CXX_MODULE_INTERNAL_PARTITIONS`: for any TU which is a partition
>>> without `export` and not an implementation unit (needs to be
>>> separate because of the `-internalPartition` flag is needed during
>>> scanning; it wouldn't be that important if the non-standard MSVC
>>> extention needed the flag and this one did not)
>> IIUC these are what the std calls implementation partitions. It is
>> unfortunate the same name is not being used.
> I'd also prefer this name as coined by the standard instead of a
> tool-specific one.
>
>
> _______________________________________________
> SG15 mailing list
> SG15_at_[hidden]
> https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.isocpp.org%2Fmailman%2Flistinfo.cgi%2Fsg15&data=05%7C01%7Cgdr%40microsoft.com%7Cbc42e15e539f4b71ecbb08da2dc0f3a4%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C637872603993970460%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=PP6%2FjkGQIKfh1FJPFy47qlu8m04El5vgEO8Z7dTz5Uo%3D&reserved=0
Received on 2022-05-04 11:45:13