[llvm] r174340 - Added LLVM Asm/Bitcode Reader/Writer support for new IR keyword externally_initialized.

[Michael Gottesman]

This has been extensively discussed. See:

http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20130107/161873.html
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20130114/162024.html
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20130114/162064.html
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20130114/162090.html
http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20130114/162221.html

*NOTE* This is just one large email chain that the archives broke up for some reason. I wanted to make sure that you could see it all so I just included links to the top parts of what the archives thought was a thread.

Michael

On Feb 5, 2013, at 10:59 AM, Duncan Sands <baldrick at free.fr> wrote:

> Hi Michael,
> 
> On 05/02/13 19:44, Michael Gottesman wrote:
>> Yes. See r174270. I was trying to be incremental (perhaps too incremental = p).
> 
> thanks, I found it now.  I hate to say this but it seems like a rather dubious
> concept to me.  Presumably what you want is an "internal" variable, so all kinds
> of optimizations can happen, but you don't want too many optimizations to
> happen because in fact external places can and do touch it.  Your approach
> sounds a lot like lying to the optimizers to me, which is usually a bad idea
> (often tempting though!).  What is the situation that provoked all this?
> 
> Ciao, Duncan.
> 
>> 
>> Michael
>> 
>> On Feb 5, 2013, at 6:43 AM, Duncan Sands <baldrick at free.fr> wrote:
>> 
>>> Hi Michael,
>>> 
>>>> Added LLVM Asm/Bitcode Reader/Writer support for new IR keyword externally_initialized.
>>> 
>>> did you also document it somewhere?
>>> 
>>> Ciao, Duncan.
>>> _______________________________________________
>>> llvm-commits mailing list
>>> llvm-commits at cs.uiuc.edu
>>> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>> 
>