[PATCH] D20260: IR: Introduce local_unnamed_addr attribute.

[Rafael Espíndola via llvm-commits]

> Yes, there's a patch, and it explains what's happening as "Minimise the
> diff a little" and "Add assembler/bitcode test". The description of
> what's happening is completely missing. The comments so far aren't on
> the mailing list at all, and somebody who reads this mail first has no
> idea what's going on. Is the patch something interesting? Should I spend
> my time following the link and catching up? It also has a subject that
> starts with "Re:", which makes me and my email filter think I've already
> read the first patch and decided whether or not to pay attention.
>
> Essentially, patches that start without llvm commits don't look like
> mails that need to be read.
>
> The first email in a patch review thread is incredibly important. It
> summarizes the point of the patch, provides context, and has the patch.
> Coming into the conversation half way through is very difficult.
>
>> That said the patch is always a mere two-clicks away from the email: http://
>> reviews.llvm.org/D20260?download=true
>> One could even write a plugin for his email client to act on link in the email
>> "http://reviews.llvm.org/D20260" and automatically curl the latest patch.
>
> How do I know if I need or want to look at it? Is "Minimise the diff a
> little" an interesting patch?

If there is information missing on a thread, reply to that thread
asking that more information be added. That is true regardless of what
the thread is or how it was created. A thread that started with an
incomplete phab setup is just a special case.

Cheers,
Rafael