On 4/14/2017 7:14 AM, Martell Malone wrote:
> Updated clang no longer defines __IA32INTRIN_H so lets do this properly.
It appears that the only use for in that file is the
prototype for __rdtsc. While it's not "good form" to have the prototype
in multiple places...
> I assume intrin-impl.h
> We have text/x-patch, not application/x-patch. I don't understand the
> difference. Do you?
The problem with mime types is that anyone can use any string they want
and it means whatever they want it to mean. Of course to be "generally
useful" the meaning must be "generally agreed upon."
> I don't think application/octet-stream is something we would want to
> enable. Doesn't that sound like something that would be abused?
I suppose application/octet-stream might be useful if we commonly worked
with binary files for patches. But I agree with you, don't do it.
A google search
On 2/9/2017 3:28 AM, Jacek Caban wrote:
> __has_builtin looks like a nice way to fix it.
Too bad gcc doesn't see the value in __has_builtin:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=66970
dw
--
Check out the
Hmm.
It seems a bit backwards to have the function that takes a 'long double'
calling the function that takes a 'double.' Yes, they are both the same
size on ARM, but I think I would have gone the other way. Plus I kinda
like having all the implementations in one file (fmal.c).
Other than
So you have decided that __builtins can't be used then? That's too bad.
I know almost nothing about the guts of floating point, so I'm prepared
to defer to your judgement, but here's what I think:
Let me propose an alternative for fma.c:
/**
* This file has no copyright assigned
On 1/18/2017 5:14 AM, lhmouse wrote:
> Patch is attached.
I see that you have replaced the x86 parts for fma and fmaf with C
code. That seems like a good thing. Is there some reason you can't do
that with the ARM versions too?
Reducing the amount of platform-specific code also seems like a
; fields, their code would break.
While I like this "cleaner" approach, playing it safe means leaving this
alone.
Clean? Or safe?
dw
On 12/25/2016 9:49 PM, David Wohlferd wrote:
OK, clear to apply to master, thanks for testing!
When I went to push, I had a problem with git tha
OK, clear to apply to master, thanks for testing!
When I went to push, I had a problem with git that took me a while to
sort out. While I was doing that, I realized that this patch wasn't
quite right (so I never push-ed it). This x87 stuff is just weird.
According to 7.6.4.2 in c99,
On 12/24/2016 8:36 AM, Michele Bucca wrote:
> ../../../gcc-6.2.0/libgcc/libgcc2.c:557:1: warning: control reaches
> end of non-void function [-Wreturn-type]
What does your gcc configure line look like? Have you tried adding
--enable-werror=no?
dw
On 12/21/2016 4:27 AM, JonY wrote:
> On 12/21/2016 07:29 AM, David Wohlferd wrote:
>> Comments? Zidane, can you try this?
> David, can you also test against the gfortran testsuites?
Since mingw-w64 doesn't have any gfortran testsuites, I assume you want
me to run gcc's. Running &q
fix is something like the attached. This undoes
kai's change, and fixes feholdexcept instead. I've taken a brief look,
and I don't see any other functions that are like this.
Comments? Zidane, can you try this?
dw
On 12/19/2016 7:03 PM, David Wohlferd wrote:
FWIW, I agree that this code
FWIW, I agree that this code does not look right.
The purpose of fesetenv(const fenv_t * envp) is to change the settings
of the floating point environment as specified by the parameter.
However, instead of using the values from the parameter, the current
code *overwrites* the passed in
Apparently there is a filter on the SF mail list. Nightstrike had to
specifically add something for my patches to get thru (see
https://sourceforge.net/p/mingw-w64/mailman/message/35277894/).
dw
On 11/11/2016 2:30 AM, JonY wrote:
> On 11/11/2016 18:17, LRN wrote:
>> On 11.11.2016 13:09, JonY
Is there an official list of release criteria for mingw-w64 v6? Or a
targeted release date? I'm not seeing anything on the web site.
I have some changes I'd like to propose that only belong at the start of
a new cycle. But I'd like to see how they fit with the current plans.
dw
> RC has been out forever without much feedback.
Ahh. I did not know this.
Looking back thru the archive, I see the announcements came during a
time (Feb 25 & Mar 25) when I was not following the list. I didn't tune
back in until July.
Of course this does raise another obvious question
On 10/18/2016 3:28 PM, JonY wrote:
> -BEGIN PGP SIGNED MESSAGE-
> Hash: SHA256
>
> On 10/19/2016 04:18, Jean-Baptiste Kempf wrote:
>> On 18 Oct, David Wohlferd wrote :
>>> Is there an official list of release criteria for mingw-w64 v5?
>>> Or a ta
Is there an official list of release criteria for mingw-w64 v5? Or a
targeted release date? I'm not seeing anything on the web site.
I have my own list of things I'm working on (~6 items) that I'd like to
see included, assuming there is still time. I also have a few more that
I'd like to do,
On 10/17/2016 2:34 AM, JonY wrote:
> On 10/17/2016 10:17, David Wohlferd wrote:
>> Is that the same as "approved for push?" Or am I still waiting to
>> hear from someone? I'm told you are the 'makefile' approver.
>>
> Yes, you can go ahead
On 10/15/2016 5:04 PM, JonY wrote:
> -BEGIN PGP SIGNED MESSAGE-
> Hash: SHA256
>
> On 10/16/2016 07:21, David Wohlferd wrote:
>> I'm asking that my current patch (which does NOT use the beta
>> files) be approved. If there is a reason to use the beta files,
>&g
I have a patch to bring all the mingw-w64 build files up to the latest
version (1.15). Due to size, it is at
http://www.LimeGreenSocks.com/gen2.7z and is unchanged since I posted it
back on Oct 2. This patch is 100% regenerated files from running
"autoreconf -fiv" in all the directories.
On 10/8/2016 3:09 AM, Adrien Nader wrote:
> Yet, it would also be useful to know which compilers are used or we
> are bound to repeatedly break them. Please, report in this thread if
> you use something else than GCC or Clang.
Seems like such an obvious question, doesn't it? But I've
On 10/2/2016 6:38 PM, JonY wrote:
On 10/3/2016 06:37, David Wohlferd wrote:
So, my testing didn't turn up any problems. The patch is pretty
big (1,389,398), so I have compressed it and uploaded it to
http://www.LimeGreenSocks.com/gen2.7z (where it is only 82,083).
Just a reminder: Despite
On 10/1/2016 4:37 PM, David Wohlferd wrote:
> On 9/24/2016 7:23 PM, David Wohlferd wrote:
>> On 9/24/2016 4:55 AM, NightStrike wrote:
>>> On Sep 23, 2016 8:08 PM, "David Wohlferd" <d...@limegreensocks.com>
>>> wrote:
>>>> Ok. With that
On 9/24/2016 7:23 PM, David Wohlferd wrote:
> On 9/24/2016 4:55 AM, NightStrike wrote:
>> On Sep 23, 2016 8:08 PM, "David Wohlferd" <d...@limegreensocks.com> wrote:
>>> Ok. With that change in place, I have run
>>>autoreconf -fiv
somewhere after rev 4680 (b028d1)
dated 2016-07-08 and 4721 (1006bb) dated 2016-08-22.
None of *uuid.c source files changed this year so it's even more strange.
crt-git package and it's sources can be found here
http://repo.msys2.org/mingw/
2016-09-30 6:58 GMT+02:00 David Wohlferd &l
Just those few? How about the rest of extras-uuid.c? The only ones
that don't look like dupes are:
DEFINE_GUID(IID_IEnumSTATURL,0x3c374a42,0xbae4,0x11cf,0xbf,0x7d,0,0xaa,0,0x69,0x46,0xee);
DEFINE_GUID(IID_IHttpNegotiate,0x79eac9d2,0xbaf9,0x11ce,0x8c,0x82,0,0xaa,0,0x4b,0xa9,0xb);
Fixes this warning which occurs when wchar_t is used:
.../mingw-w64-crt/misc/dirent.c:121:3: warning: 'memset' used with
length equal to number of elements without multiplication by element
size [-Wmemset-elt-size]
memset (nd->dd_dir.d_name, 0, 260 /*FILENAME_MAX*/);
Push?
dw
diff --git
On 9/24/2016 4:55 AM, NightStrike wrote:
> On Sep 23, 2016 8:08 PM, "David Wohlferd" <d...@limegreensocks.com> wrote:
>> On 9/23/2016 5:08 AM, JonY wrote:
>>> On 9/23/2016 09:50, David Wohlferd wrote:
>>>> On 9/22/2016 5:53 AM, JonY wrote:
>>&g
On 9/22/2016 5:53 AM, JonY wrote:
> On 9/22/2016 20:23, JonY wrote:
>> just remove the mpdecimal directory, the dfp/*.c stuff should
>> still work, likewise for the pformat.c changes. I'm doing just that
>> since yesterday, running build tests. SF ate my GPG signed mail.
>>
>> Use format-patch -D
On 9/22/2016 9:35 AM, Hugo Beauzée-Luyssen wrote:
> On 09/08/2016 08:23 PM, David Wohlferd wrote:
>> On 9/8/2016 10:36 AM, Hugo Beauzée-Luyssen wrote:
>>> This only happens when building with -lwindowsapp (see another patch of
>>> mine). When building with the defa
On 9/20/2016 3:39 PM, David Wohlferd wrote:
On 9/20/2016 3:01 AM, JonY wrote:
On 9/20/2016 14:29, David Wohlferd wrote:
1) The paths for the math/DFP stuff are totally messed up.
Directory names include a version number (2.3) which our tree has
removed, some directory names have changed (s/cmd
On 9/20/2016 3:01 AM, JonY wrote:
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA256
On 9/20/2016 14:29, David Wohlferd wrote:
1) The paths for the math/DFP stuff are totally messed up.
Directory names include a version number (2.3) which our tree has
removed, some directory names have changed (s
On 8/12/2016 5:34 AM, NightStrike wrote:
On Thu, Aug 11, 2016 at 2:05 AM, dw wrote:
I have been told that "autoreconf -fiv" regenerates all the files using
the current version. But after seeing the dozen files it changed and
realizing I didn't actually know what any
I assume someone still needs to approve this before I push?
dw
On 9/18/2016 6:10 AM, Mateusz wrote:
> Hello,
> While I'm not able to reply to the mailing list directly (forgot to
> subscribe). Please discard my patch and commit your new patch.
> Thank you for your help!
> Best regards,
>
I believe the values from errno.h are intended to provide POSIX
compatibility for the (now defunct) Windows posix subsystem. I'd
wouldn't expect any API from XP or later to return these values.
OTOH, I don't know how the errcclass from gcc's error_constants.h is
actually used. Maybe somebody
It took me a bit to figure out the difference between my test code
(which requires the patch) and yours (which can't have it). It was subtle.
Try this patch (attached).
dw
On 9/16/2016 1:39 AM, Mateusz wrote:
Hello,
Since this commit I'm not able to compile Qt5 with GCC 6.1.0 and 6.2.0
On 9/8/2016 10:36 AM, Hugo Beauzée-Luyssen wrote:
> This only happens when building with -lwindowsapp (see another patch of
> mine). When building with the default -lkernel32, all is good, since
> kernel32.lib contains GetStartupInfo.
> windowsapp.lib, on the other hand, doesn't; which makes sense
On 9/7/2016 2:19 PM, Hugo Beauzée-Luyssen wrote:
>> Is this to deal with the fact that __tmainCRTStartup uses it?
> Exactly. In the end, the function won't be used since __tmainCRTStartup
> won't be used for a windows store app, but we still need the configure
> scripts to be able to compile
I'm confused. Is GetStartupInfo even a winstore function? Is this to
deal with the fact that __tmainCRTStartup uses it?
dw
On 9/7/2016 1:31 AM, Hugo Beauzée-Luyssen wrote:
> On 09/07/2016 10:25 AM, Jacek Caban wrote:
>> Hi Hugo,
>>
>> On 06.09.2016 16:11, Hugo Beauzée-Luyssen wrote:
>>>
So what happened here? The winerror.h that got checked in still has the
duplicate defines?
Is there some reason I can't remove the dupes (attached)?
dw
On 8/24/2016 9:46 AM, Jacek Caban wrote:
Hi Ruben,
I'm sorry I didn't look at this earlier.
I committed winerror.h (it seems that you
There is an entire collection of routines in mingw-w64-crt\secapi.
These functions replicate the functionality of a number of secure
versions of functions (_vcprintf_s, _wstrtime_s, etc). The idea was to
support these functions on platforms which didn't have the necessary DLLs.
The purpose
In mingw-w64-headers/crt/time.h there is some code that does:
#ifndef _WTIME_DEFINED
#define _WTIME_DEFINED
After that, it defines a number of functions.
mingw-w64-headers/crt/wchar.h has this same #if block, using that exact
same define, but it defines fewer functions. So if you include
The thing that started the _SECIMP work was a whole bunch of "redeclared
without dllimport attribute: previous dllimport ignored" warnings.
While the _SECIMP work (coming soon) will fix most of them, there are a
few others that still give this warning.
This patch fixes the places that DON'T
Use correct prototype for _difftime32 & _difftime64.
Note that time.h already uses the correct prototype.
Question: Could the intent here have been to treat the (normally) signed
__time32_t as unsigned in an attempt to squeeze more life out of it (the
'2038' problem)? If so, 'tricking' the
Let's try this again. This time the proposed patch is attached which
may help. For 'ease of review,' this patch does not include the
'generated' makefile.in files.
The problem I am trying to fix is that when building mingw-w64, the
compiler will often use headers from the Tools Directories
On 8/18/2016 11:27 PM, David Wohlferd wrote:
> My next patch is very small, but it affects a bunch of code. Ponder
> it a bit before approving. The goal is to fix all the warnings like
> this:
>
> warning: '_unlock_file' redeclared without dllimport attribute:
> previous
On 8/21/2016 11:19 PM, Martin Storsjö wrote:
> On Sun, 21 Aug 2016, David Wohlferd wrote:
>
>> By definition, functions with variable numbers of parameters cannot be
>> stdcall. Clang complains (warning: stdcall calling convention ignored on
>> variadic function).
>>
On 8/21/2016 11:27 PM, Martin Storsjö wrote:
> On Sun, 21 Aug 2016, David Wohlferd wrote:
>
>> Under certain circumstances, the #pragma pop_macro("__has_builtin") at the
>> bottom of intrin-impl.h can be called without ever having hit the #pragma
>> push_macro(&
On 8/21/2016 11:17 PM, Martin Storsjö wrote:
> On Sun, 21 Aug 2016, David Wohlferd wrote:
>
>> To my surprise, these two statements have (slightly) different meanings:
>>
>> STDAPI MFUnregisterPlatformFromMMCSS ();
>> STDAPI MFUnregisterPlatformFromMMCSS (void);
&
Under certain circumstances, the #pragma pop_macro("__has_builtin") at
the bottom of intrin-impl.h can be called without ever having hit the
#pragma push_macro("__has_builtin") at the top. Clang warns about this,
so I have moved the push appropriately.
Patch attached.
dw
diff --git
To my surprise, these two statements have (slightly) different meanings:
STDAPI MFUnregisterPlatformFromMMCSS ();
STDAPI MFUnregisterPlatformFromMMCSS (void);
And clang complains about it (warning: function with no prototype cannot
use the stdcall calling convention).
I have added 'void'
In this function:
long double sinhl(long double x)
there is a call to fabs:
(fabs (x) > (MAXLOGL + LOGE2L)))
However, fabs doesn't take a (long double), it only takes a (double).
Clang complains about the truncation (warning: absolute value function
'fabs' given an argument of
To get this error, I have to have a line like:
#define _FILE_OFFSET_BITS
_FILE_OFFSET_BITS should either be undefined, or be set to 32 or 64 (see
https://www.gnu.org/software/libc/manual/html_node/Feature-Test-Macros.html).
'Blank' is not a valid setting.
I'm not sure where this is happening
My next patch is very small, but it affects a bunch of code. Ponder it
a bit before approving. The goal is to fix all the warnings like this:
warning: '_unlock_file' redeclared without dllimport attribute:
previous dllimport ignored [-Wattributes]
This occurs because our headers use:
Most people could probably check this patch in "as obvious." But I
don't really speak script, so someone better check.
BTW, don't bother trying to build --enable-warnings=4 (or higher). Even
with this patch, it's not going to build (not even close).
dw
diff --git a/mingw-w64-crt/configure
It's late, so just 3 quick new ones tonight:
Ok, kai signed off on these too.
So, next are the IDL files. I (mistakenly) tried to regenerate my
changes by just running widl against the appropriate idl file.
Apparently the correct way is to use --with-widl on the configure line.
So I did
On 8/18/2016 4:23 AM, Martell Malone wrote:
> I would also like to point you to my selectany patch.
> You asked why that wasn't supported on clang on irc.
> https://github.com/martell/mingw-w64-clang/blob/master/patches/clang/0003-mingw-w64-enable-support-for-__declspec-selectany.patch
> This
I assume I still need to wait for someone else to approve this before
I push?
Kai signed off on these via IRC.
It's late, so just 3 quick new ones tonight:
e_pow.patch - Signed/unsigned compare
mingw_pformat.patch - Don't use feature (__attribute__((gcc_struct)))
that isn't supported on
Huh. Well, that went a lot better than I thought it would...
On 8/17/2016 1:11 AM, Kai Tietz wrote:
> Hello dw,
>
> 2016-08-17 4:22 GMT+02:00 David Wohlferd <d...@limegreensocks.com>:
>> One of my upcoming patches is going to be a bit controversial (or maybe
>>
One of my upcoming patches is going to be a bit controversial (or maybe
I'm just missing something). In either case, I'm going to go ahead and
start the discussion early.
First a bit of vocabulary. I'm not including this because I think
people here don't know these terms. It's because *I*
On 8/16/2016 7:30 AM, Martin Storsjö wrote:
> Btw,
>
> All of your mails end up sorted in the spam box by gmail (at least for
> me), citing this reason: "It has a from address in yahoo.com but has
> failed yahoo.com's required tests for authentication."
>
> As long as I remember to check the spam
Sorry for the spam, but if this list is how patches are expected to be
sent, I've got to get this working. I'm using gmail's web interface
instead of TB this time. I'm also only sending 1 patch. Fingers crossed...
ext2.patch - Fix macro pasting error and duplicate symbol names.
dw
Can someone who knows more about clang than I do help me out here?
snow_xmas posted a problem report to the SF forum regarding using clang with
mingw-W64. After a brief investigation, it became clear that the problem was
that clang doesn't support one of the features of inline asm that gcc
64 matches
Mail list logo