[M3devel] missed the memo? LINUXLIBC6 building WinBase?
Dragiša Durić
dragisha at m3w.org
Sat Nov 23 23:32:39 CET 2013
I think some interim release, of “get on with Modula-3 fast” kind, is better idea than to slow progress. And I don’t dislike this current state.
As for sommunity shrinking… Maybe some web facelift would be good idea… Anybody willing?
dd
On 23 Nov 2013, at 21:33, Rodney M. Bates <rodney_bates at lcwb.coop> wrote:
>
>
> On 11/23/2013 05:13 AM, Jay K wrote:
>> This is deliberate, and slightly useful, and the warnings shouldn't occur in the head,
>> for several years now. Please upgrade more frequently.
>>
>>
>
> I feel very strongly that the latest release compiler should always build the head.
> We have all too few Modula-3 users anyway. If we don''t have this ability,
> then we become an exclusive club with unreasonable barriers to outsiders.
> There is hardly a better way to keep the community shrinking.
>
> I know it can be a pain to maintain compatibility, but we really must do it.
> Until we get a new release with a compiler that won't choke on WINAPI, we
> need to keep different versions of the source.
>
> Actually, I thought I had removed all occurrences of this problem a few
> months ago. How did I miss this one?
>
>
>> In particular:
>>
>>
>> - It is useful because "much" of the interfaces are describing types,
>> and types aren't just for function parameters, for functions that
>> aren't available, but they are used as part of files, and you can,
>> hypothetically, use them in your own functions (heck, you can
>> write a Win32-on-Posix compatibility layer and provide the functions..).
>>
>>
>> In particular, mklib uses these types and is fairly portable.
>> It might have endian/alignment problems though, esp. endian.
>>
>>
>> I do have a plan to get rid of mklib though...i was slow to realize it
>> because mklib exports too much and I wanted a replacement wthout that flaw,
>> but just matching it is adequate. The frontend/backends shall output
>> the .def files instead of mklib. And then the usual lib/ar/link can/should
>> make the libraries. If anyone wants to use LTCG on NT with the C backend,
>> this is the way (or put everything in one source file via the backend, not a bad idea)..
>>
>>
>> - In the past 20 years or so, there is only one widely used
>> platform that made the mistake of having multiple calling conventions
>> and that is NT/x86. Not NT in general, but NT/x86 specifically.
>> I expect nobody will make this mistake again.
>> I'm not sure MIPSo32/MIPSn32 or ARM-mess count as the same mistake.
>> On NT/x86, the calling convention does not bifurcate the entire target
>> and force building everything multiple times, because you can sprinkle it around and
>> each "type" can call each "type". This is good and bad.
>> The other OS/processor combinations with multiple calling conventions
>> define it as multiple ABIs with I believe no offical interop (except at the
>> kernel level, for some of them).
>>
>>
>> Therefore, callling conventions can be safely ignored on all platforms except NT/x86.
>> Nobody will have any names to use here, overlapping the NT/x86 names or not.
>> If the day comes where I am wrong and some target has multiple conventions,
>> we should change the pragma to accept multiple. If the names overlap with NT/x86,
>> we would put platform qualifiers on them. Never again should we have the Win32 SQL.i3
>> and Posix SQL.i3 that are the same, except for calling convention. What a waste -- almost
>> the same, but for one little difference that could be handled another way, they go duplicated.
>> That pattern bugs me.
>>
>>
>> - Jay
>>
>>
>>
>>
>> ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
> --
>> From: dragisha at m3w.org
>> Date: Fri, 22 Nov 2013 23:55:23 +0100
>> To: m3devel at elegosoft.com
>> Subject: Re: [M3devel] missed the memo? LINUXLIBC6 building WinBase?
>>
>> Yes, it was in a memo. Sorry for disruption :).
>>
>> On 22 Nov 2013, at 23:31, Dragiša Durić <dragisha at m3w.org <mailto:dragisha at m3w.org>> wrote:
>>
>> + ./scripts/do-pkg.sh buildship m3core libm3 tcp
>> /root/rpmbuild/BUILD/cm3/scripts/pkgmap.sh -c cm3 -build -DROOT='/root/rpmbuild/BUILD/cm3' && cm3 -ship -DROOT='/root/rpmbuild/BUILD/cm3' m3core libm3 tc
>> === package m3-libs/m3core ===
>> +++ cm3 -build -DROOT='/root/rpmbuild/BUILD/cm3' $RARGS && cm3 -ship $RARGS -DROOT='/root/rpmbuild/BUILD/cm3' +++
>> --- building in LINUXLIBC6 ---
>>
>> ignoring ../src/m3overrides
>>
>> new source -> compiling WinBase.i3
>> "../src/win32/WinDef.i3", line 131: warning: unrecognized pragma (ignored) (WINAPI)
>> "../src/win32/WinDef.i3", line 132: warning: unrecognized pragma (ignored) (WINAPI)
>> "../src/win32/WinDef.i3", line 133: warning: unrecognized pragma (ignored) (WINAPI)
>> "../src/win32/WinBase.i3", line 198: warning: unrecognized pragma (ignored) (WINAPI)
>> "../src/win32/WinBase.i3", line 918: unsupported language or calling convention (WINAPI)
>> "../src/win32/WinBase.i3", line 921: unsupported language or calling convention (WINAPI)
>> "../src/win32/WinBase.i3", line 924: unsupported language or calling convention (WINAPI)
>> "../src/win32/WinBase.i3", line 927: unsupported language or calling convention (WINAPI)
>> "../src/win32/WinBase.i3", line 930: unsupported language or calling convention (WINAPI)
>> "../src/win32/WinBase.i3", line 933: unsupported language or calling convention (WINAPI)
>> "../src/win32/WinBase.i3", line 943: unsupported language or calling convention (WINAPI)
>> "../src/win32/WinBase.i3", line 949: unsupported language or calling convention (WINAPI)
>> "../src/win32/WinBase.i3", line 952: unsupported language or calling convention (WINAPI)
>> "../src/win32/WinBase.i3", line 955: unsupported language or calling convention (WINAPI)
>>
>> --
>> Dragiša Durić
>> dragisha at m3w.org <mailto:dragisha at m3w.org>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://m3lists.elegosoft.com/pipermail/m3devel/attachments/20131123/5a0bd6dd/attachment-0002.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 495 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <http://m3lists.elegosoft.com/pipermail/m3devel/attachments/20131123/5a0bd6dd/attachment-0002.sig>
More information about the M3devel
mailing list