[LLVMdev] Final Windows MinGW buildbot slave document...

[Mikael Lyngvig]

Thanks for your feedback!

Yes, splitting the document makes perfect sense.

I'll revise the document and redo it in Sphinx (am curious about that one).
 It is using the LLVM CSS, though.  You just need to make sure the CSS file
is in the same directory as the document - it even uses the lines.gif image
(from the img subdirectory of the current folder).

As for stable GCC, I bet that's something about cross-building GCC/Win64 on
Linux or another Unix?  I picked the Drangon release because it is
off-the-shelf and seems to work well.  But I guess I can figure out how to
build a stable GCC release eventually.

Basically, you have to disable the antivirus solution because of a single
file in the ClamAVS files that are part of LLVM test suite.  I ran into
this when I first tried to checkout the sources: the checkout was aborted
because even Microsoft Security Essentials (the worst AV solution on the
market) managed to detect a virus in one of the files.  If we could get rid
of that file, there'd be no need to disable AV.  I'll try doing a checkout
later and find the offending file and submit a bug report.  Then others can
decide what to do.

2012/6/11 Justin Holewinski <justin.holewinski at gmail.com>

> On Sat, Jun 9, 2012 at 12:40 AM, Mikael Lyngvig <mikael at lyngvig.org>wrote:
>
>> Hi,
>>
>> Here's the final, confirmed working version of my Windows guide for how
>> to set up a MinGW buildbot slave.  Please consider to include it in the
>> documentation on the LLVM website.
>>
>> Basically, it saves the newcoming Windows user about two or three days of
>> experimentation to get it all working.  I think those days are important as
>> many people balk at spending days on figuring out how to get flying when
>> all they want is to take a look at LLVM and then perhaps later on become
>> more and more involved with the dev group.
>>
>> If you don't like the document and/or won't use it, I won't be saddened.
>>  You can sanely argue that the document is redundant, and you can also
>> argue that it is very helpful. Both views are equally valid as far as I am
>> concerned.  I just happen to like documentation that spells out what needs
>> to be done because we're not all gurus from day one.
>>
>> I am sending it to this list because I don't know where else to post it.
>>  llvm-commits seems a bit overkill for a document.
>>
>>
> Some quick comments:
>
>    1. Some of the points against Visual Studio are not necessarily true,
>    e.g. there is a free version, and a lot of non-Microsoft tools/technologies
>    work with it.
>    2. This may be worth splitting into two documents: (1) building Clang
>    with MinGW on Windows, and (2) creating a MinGW-based Clang build-bot.
>    3. Since this is a new document, it may be worth writing it in
>    Sphinx/ReST and integrating it with the ReST files under docs/.  Or at the
>    very least, use the LLVM CSS style to match the existing HTML style.
>    4. I'm sure about advocating for disabling of anti-virus during the
>    build.
>    5. Why tie this to the drangon MinGW64 release?  It would be nice to
>    make sure it also works for a stable GCC release.
>
> This is by no means a thorough critique, just a few points that jumped out
> at me.  Please re-submit this as a patch to llvm-commits, so it can go
> through the proper review channels.
>
>
>>
>> Cheers,
>> Mikael
>> -- Love Thy Frog!
>>
>>
>> _______________________________________________
>> LLVM Developers mailing list
>> LLVMdev at cs.uiuc.edu         http://llvm.cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev
>>
>>
>
>
> --
>
> Thanks,
>
> Justin Holewinski
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20120611/2c41251a/attachment.html>