similar to: [LLVMdev] RE: MinGW Tablegen

Hi all, Well, suggestion for workarounds for the namespace problems are welcome... this is a 7 minutes compile files on a pentium 4 3ghz 700Mb ram... The fatal error at the end MAY depend on the previous... or at least, I hope so. cl /nologo /TP /EHsc /GR /Zi /Yd /D__STDC_LIMIT_MACROS /DHAVE__FINITE_IN_FLOAT_H /DHAVE__ISNAN_IN_FLOAT_H /ISTLport-4.6.2\stlport /Illvm\inc lude

struct X86AsmPrinter is in an anonymous namespace, but printInstruction is declared in namespace llvm. try editing the tablegen output to move X86AsmPrinter::printInstruction into an anonymous namespace, not llvm. I suspect this will at least fix the first problem. Then we can figure out the proper longterm fix. Andrew On Tue, 2004-10-12 at 03:56, Paolo Invernizzi wrote: > Hi all, >

Some problems... (or I'm missing something...) 1) asmwriternum seems to be not supported... scons: Building targets ... d:/home/arathorn/sandbox/llvm/tblgen.exe -gen-asm-writer -asmwriternum=1 -I llvm\lib\Target\X86 -o tablegen_includes\X86GenIntelAsmWriter.inc llvm\lib\Target\X86\X86.td llvm\lib\Target\X86\X86InstrInfo.td llvm\lib\Target\X86\X86RegisterInfo.td llvm\lib\Target\Target.td

It would be great to avoid STLPort and use plain vanilla VC... as I told, the biggest difference it's how the hash_map and hash_set are implemented, but I'm not so strong in C++ for resolving the iussue. About the build procedure, it's based on scons, and it's still at a very preliminary stage... Right now I'm trying to build TableGen with it, as till now I've always

I'll wait for the research. We should try, as much as possible, to make it work with just what the compiler provides and without third party packages. Thanks, reid. On Fri, 2004-09-24 at 07:46, Jeff Cohen wrote: > I added the include of cstudio and it fails with plain VC7.1; the file > does not exist. > > Add it for now. If it is impossible to build with VC7.1 and without

Uh... this may be a silly question, but why can't you include <stdio.h>? It'd be much better than <iostream>. Anyway, I think I'll try this weekend to come up with my own way of building on Win32. I prefer that building on Windows depends only on Microsoft and GNU tools, and the fewer of the latter the better. My gut instinct is to capture all the files generated by

>From: Paolo Invernizzi <arathorn at fastwebnet.it> >Date: Thu, 16 Sep 2004 10:20:39 +0200 > >I'm using scons to generate that files from .in files. I implemented in it >the configure check regarding iterators, hash and so on... >something like: > Hey, you've found the tool that makes it possible to generically reading Makefiles... Cool - The tool I've

But I compiled that under vc7.1 as it was! On Fri, 24 Sep 2004 15:19:22 +0200 Paolo Invernizzi <arathorn at fastwebnet.it> wrote: > Adding an include for std::remove under vc7.1 > > --- > Paolo Invernizzi >

Well, I just tried to copy the files generated by configure to Windows for building there. It's a non starter. They are totally dependent on the Unix environment and it would be simpler to start from scratch. The scons approach of generating VC project files is very interesting (but what about the solution file?). Unfortunately, a serious attempt at this would take far more time than I

Adding an include for std::remove under vc7.1 --- Paolo Invernizzi -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: diff.txt URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20040924/e1ca1218/attachment.txt>

Hi all, I cannot figure out why is named GenRegisterInfo.h.inc and not GenRegisterInfo.inc ... Is it for a dependency problem? Back again to compilation problems under win32 with VC llvm\lib\Analysis\DataStructure\Local.cpp(628) : error C2105: '--' needs l-value the line is: Result.mergeWith(getValueDest(**--CS.arg_end())); Can I submit patches for mutate it in something like:

<algorithm> works too. On Fri, 24 Sep 2004 10:09:21 -0500 Alkis Evlogimenos <alkis at cs.uiuc.edu> wrote: > On Fri, 2004-09-24 at 09:43, Paolo Invernizzi wrote: > > Jeff Cohen wrote: > > > > >But I compiled that under vc7.1 as it was! > > > > > > > > ;-(( > > > > Probably is an implicid includes, but I'm using the

I checked right now that it compiles also with #include <iostream> Jeff, can you test it with plain VC? --- Paolo Invernizzi On Sep 24, 2004, at 4:52 PM, Reid Spencer wrote: > I'll wait for the research. We should try, as much as possible, to make > it work with just what the compiler provides and without third party > packages. > > Thanks, > > reid. > >

OK. I strongly support that sentiment. Paolo, could you send me your procedure for building under Windows? I haven't tried to build anything but System/Win32 so far. On Fri, 24 Sep 2004 07:52:23 -0700 Reid Spencer <reid at x10sys.com> wrote: > I'll wait for the research. We should try, as much as possible, to make > it work with just what the compiler provides and without

MSVC++ is picky about this. It considers classes and structs to be different types so you have to be consistent. If you forward declared a struct as a class within the same compilation unit, it would complain about that too. It's not just linking. On Thu, 23 Sep 2004 15:59:42 +0200 Paolo Invernizzi <arathorn at fastwebnet.it> wrote: > Hi all, > > Finally I managed to find

Someone needs to adjudicate on whether I add the #include of <cstdio> or not. I can't test this so, Paolo/Henrik/Jeff, please let me know if I need to add it. Thanks, Reid. On Fri, 2004-09-24 at 07:08, Jeff Cohen wrote: > But I compiled that under vc7.1 as it was! > > On Fri, 24 Sep 2004 15:19:22 +0200 > Paolo Invernizzi <arathorn at fastwebnet.it> wrote: > >

Here is the pre-pre-pre alpha of the file, llease be kind <g> I give up on TableGen... cannot build the flex/bison emitted files ;-( With my hacked version of the checkout the script build Fibonacci.exe and HowToUseJIT.exe among with the proper libraries. I included also a demo version of the HowToUseJIT Visual studio project generated by the script. You can debug the program in the

On Sep 23, 2004, at 4:08 PM, Jeff Cohen wrote: > MSVC++ is picky about this. It considers classes and structs to be > different types so you have to be consistent. If you forward declared > a > struct as a class within the same compilation unit, it would complain > about that too. It's not just linking. You are right... BTW, I've just fixed that problem in my checkout

I have just committed a change to Value.h that changes the Value class from using a "struct" declaration to a "class" declaration. I'm not sure why VC7.1 would generate different symbols for class vs. struct. I'm pretty certain that's a violation of the ABI. In any event, we should be consistent. The Value class is declared "class Value" in numerous places

On Tue, 12 Oct 2004, Paolo Invernizzi wrote: > Hi all, > I cannot figure out why is named GenRegisterInfo.h.inc and not > GenRegisterInfo.inc ... > Is it for a dependency problem? I'm not sure what you're saying here. In the X86 backend, for example, we generate both X86GenRegisterInfo.h.inc and X86GenRegisterInfo.inc. The former is #included into X86RegisterInfo.h and the