From mboxrd@z Thu Jan 1 00:00:00 1970 Authentication-Results: mx.groups.io; dkim=missing; spf=pass (domain: redhat.com, ip: 209.132.183.28, mailfrom: lersek@redhat.com) Received: from mx1.redhat.com (mx1.redhat.com [209.132.183.28]) by groups.io with SMTP; Wed, 08 May 2019 12:50:24 -0700 Received: from smtp.corp.redhat.com (int-mx05.intmail.prod.int.phx2.redhat.com [10.5.11.15]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id AD37C308793F; Wed, 8 May 2019 19:50:23 +0000 (UTC) Received: from lacos-laptop-7.usersys.redhat.com (ovpn-120-255.rdu2.redhat.com [10.10.120.255]) by smtp.corp.redhat.com (Postfix) with ESMTP id 7F7E262671; Wed, 8 May 2019 19:50:21 +0000 (UTC) Subject: Re: [edk2-devel] [PATCH V2] BaseTools:improve code to support C files with .C suffixes To: Leif Lindholm , "Fan, ZhijuX" Cc: Andrew Fish , "devel@edk2.groups.io" , "Gao, Liming" , "Feng, Bob C" , Michael D Kinney References: <5C1A24BA-648D-4A3F-AE88-034073AB5C13@apple.com> <20190507144002.gq3rdtappd3rupmd@bivouac.eciton.net> <55FC6AA1-43B5-4557-8A17-0E3B00E3EE9A@apple.com> <20190508090926.4r4e6jqlg45yg6yp@bivouac.eciton.net> <20190508120823.urob54m7tojotq7n@bivouac.eciton.net> From: "Laszlo Ersek" Message-ID: Date: Wed, 8 May 2019 21:50:20 +0200 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.9.1 MIME-Version: 1.0 In-Reply-To: <20190508120823.urob54m7tojotq7n@bivouac.eciton.net> X-Scanned-By: MIMEDefang 2.79 on 10.5.11.15 X-Greylist: Sender IP whitelisted, not delayed by milter-greylist-4.5.16 (mx1.redhat.com [10.5.110.45]); Wed, 08 May 2019 19:50:23 +0000 (UTC) Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: 7bit On 05/08/19 14:08, Leif Lindholm wrote: > Hi Fan Zhiju, > > I understand your reasoning, but that strengthens my view that we > should leave this change out. > > Actually, could we consider *dropping* support for .C files > altogether, since we are striving to support multiple toolchains, and > the two major families of toolchains we use consider .C files to be > fundamentally different things? > > Andrew, Laszlo, Mike? I agree -- on any case-sensitive filesystem, the ".C" suffix implies the C++ language. (Minimally for gcc, not sure about the rest.) In the first place, we should never *have* a file in the edk2 tree that ends with the upper-case ".C" suffix, due to the above. (C++ is not a supported language in edk2.) And so we don't need any BaseTools support for ".C" either. If an external platform of Packages tree contains ".C" files, then those files should be renamed to ".c", in my opinion. (Sooner or later someone will try to build that tree with gcc or clang, and then stuff will break hard.) Just my 2 cents, since I was asked. Thanks Laszlo > > Best Regards, > > Leif > > On Wed, May 08, 2019 at 10:57:09AM +0000, Fan, ZhijuX wrote: >> Hi Leif, >> >> This patch is going to fix the bug in commit 05217d210e. >> this patch and commit 05217d210e is just for MSVC. It doesn't have any effect on GCC. >> this patch does not imply to compile .C as C instead of C++. >> We just found out that they build break because the.C file was in the source file, >> But the MSVC compiler allows.C files,So I fixed this bug to Change the original logic to support.C files. >> >> >> Any question, please let me know. Thanks. >> >> Best Regards >> Fan Zhiju >> >> >> >> >>> -----Original Message----- >>> From: Leif Lindholm [mailto:leif.lindholm@linaro.org] >>> Sent: Wednesday, May 8, 2019 5:09 PM >>> To: Andrew Fish >>> Cc: devel@edk2.groups.io; Fan, ZhijuX ; Gao, Liming >>> ; Feng, Bob C >>> Subject: Re: [edk2-devel] [PATCH V2] BaseTools:improve code to support C >>> files with .C suffixes >>> >>> On Tue, May 07, 2019 at 07:01:26PM -0700, Andrew Fish wrote: >>>> >>>> >>>>> On May 7, 2019, at 7:40 AM, Leif Lindholm >>> wrote: >>>>> >>>>> Hi Fan Zhiju, >>>>> >>>>> But where does the string come from that contains a .C suffix? >>>>> Is the tool internally converting things to uppercase, or is some >>>>> source file in the build incorrectly named? >>>>> >>>> >>>> Leif, >>>> >>>> Our build system defines .C as correct! I think it has been that way a very >>> long time. >>> >>> .C is valid, but at least for GCC it is equivalent to all of the other non-.c >>> options - i.e., interpret as c++. Which is why I am wondering about the case >>> that ends up with the build system internally processing this. >>> >>> If it is changed to permitting .C files to be compiled as C instead of >>> C++ (which the patch seems to imply), that sounds incorrect to me. >>> >>> / >>> Leif >>> >>>> >>> https://github.com/tianocore/edk2/blob/master/BaseTools/Conf/build_rul >>>> e.template#L109 >>>> >>>> [C-Code-File] >>>> >>>> ?.c >>>> ?.C >>>> ?.cc >>>> ?.CC >>>> ?.cpp >>>> ?.Cpp >>>> ?.CPP >>>> >>>> Thanks, >>>> >>>> Andrew Fish >>>> >>>> >>>>> I am asking because it is not clear to me whether the patch resolves >>>>> a problem or hides one. >>>>> >>>>> Best Regards, >>>>> >>>>> Leif >>>>> >>>>> On Tue, May 07, 2019 at 03:05:02AM +0000, Fan, ZhijuX wrote: >>>>>> This problem has nothing to do with the file system, We just use >>>>>> the filename as a string to compare with other strings Our unittest >>>>>> tested minplatform, Ovmf. This problem was found when building a >>>>>> platform inside Intel. >>>>>> We've tested it on Linux and Windows. >>>>>> >>>>>> Any question, please let me know. Thanks. >>>>>> >>>>>> Best Regards >>>>>> Fan Zhiju >>>>>> >>>>>> -----Original Message----- >>>>>> From: afish@apple.com [mailto:afish@apple.com] >>>>>> Sent: Tuesday, May 7, 2019 10:31 AM >>>>>> To: devel@edk2.groups.io; Fan, ZhijuX >>>>>> Cc: Gao, Liming ; Feng, Bob C >>>>>> >>>>>> Subject: Re: [edk2-devel] [PATCH V2] BaseTools:improve code to >>>>>> support C files with .C suffixes >>>>>> >>>>>> This brings up a question? Do we tests on a file system that is case >>> sensitive? Is this just lack of a test case? >>>>>> >>>>>> Thanks, >>>>>> >>>>>> Andrew Fish >>>>>> >>>>>>> On May 6, 2019, at 7:22 PM, Fan, ZhijuX wrote: >>>>>>> >>>>>>> BZ: https://bugzilla.tianocore.org/show_bug.cgi?id=1773 >>>>>>> >>>>>>> Build break if C file suffixes of named .C instead of .c Code not >>>>>>> recognize filenames with .C suffixes. >>>>>>> >>>>>>> This patch adds code to Support both .c file and .C file >>>>>>> >>>>>>> Cc: Bob Feng >>>>>>> Cc: Liming Gao >>>>>>> Signed-off-by: Zhiju.Fan >>>>>>> --- >>>>>>> BaseTools/Source/Python/AutoGen/GenMake.py | 3 ++- >>>>>>> 1 file changed, 2 insertions(+), 1 deletion(-) >>>>>>> >>>>>>> diff --git a/BaseTools/Source/Python/AutoGen/GenMake.py >>>>>>> b/BaseTools/Source/Python/AutoGen/GenMake.py >>>>>>> index 0e0f9fd9b0..858ddedf8e 100644 >>>>>>> --- a/BaseTools/Source/Python/AutoGen/GenMake.py >>>>>>> +++ b/BaseTools/Source/Python/AutoGen/GenMake.py >>>>>>> @@ -1035,7 +1035,8 @@ cleanlib: >>>>>>> CmdTargetDict[CmdSign] = "%s %s" % >>> (CmdTargetDict[CmdSign], SingleCommandList[-1]) >>>>>>> Index = CommandList.index(Item) >>>>>>> CommandList.pop(Index) >>>>>>> - if SingleCommandList[-1].endswith("%s%s.c" % (TAB_SLASH, >>> CmdSumDict[CmdSign.lstrip('/Fo').rsplit(TAB_SLASH, 1)[0]])): >>>>>>> + if SingleCommandList[-1].endswith("%s%s.c" % >>> (TAB_SLASH, CmdSumDict[T.Target.SubDir])) or \ >>>>>>> + SingleCommandList[-1].endswith("%s%s.C" % >>> (TAB_SLASH, CmdSumDict[T.Target.SubDir])): >>>>>>> Cpplist = CmdCppDict[T.Target.SubDir] >>>>>>> Cpplist.insert(0, '$(OBJLIST_%d): $(COMMON_DEPS)' % >>> list(self.ObjTargetDict.keys()).index(T.Target.SubDir)) >>>>>>> T.Commands[Index] = '%s\n\t%s' % (' >>>>>>> \\\n\t'.join(Cpplist), CmdTargetDict[CmdSign]) >>>>>>> -- >>>>>>> 2.14.1.windows.1 >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> >>>>>> >>>>>> >>>>>> >>>>>> >>>>> >>>>> >>>>