llvm dev - Mar 2016 - Need help with code generation

On Mar 22, 2016 8:03 AM, "James Molloy" <james at
jamesmolloy.co.uk> wrote:
>
> Rafael,
>
> I appreciate that your goals are to create a performant linker, quickly.
I understand that and wouldn't want to slow you down unnecessarily,
especially as I appreciate that I'm on the sidelines and not producing
patches for LLD at the moment.
>
> I also appreciate that LLD/ELF development has certainly picked up pace
recently and we might actually have a functioning linker sometime soon.
That is very exciting, and thankyou for all your work.
>
> My worry is that the technical debt accrued will have to be paid off by
someone at some point. At some point, someone will have to come along and
make LLD robust ("enough", for some definition), and the current
decision
will make that person's job much harder.

Again: users *today* are less likely to see a crash in lld than llvm or
clang and those are much larger code bases.

Cheers,
Rafael
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20160322/dbb4618a/attachment.html>

Hi,

That's not the point at all. There are known unknowns and unknown unknowns;
creating known problems and then not at least marking them as such affects
people down the line.

Cheers,

James

On Tue, 22 Mar 2016 at 12:16 Rafael Espíndola <rafael.espindola at
gmail.com>
wrote:
>
> On Mar 22, 2016 8:03 AM, "James Molloy" <james at
jamesmolloy.co.uk> wrote:
> >
> > Rafael,
> >
> > I appreciate that your goals are to create a performant linker,
quickly.
> I understand that and wouldn't want to slow you down unnecessarily,
> especially as I appreciate that I'm on the sidelines and not producing
> patches for LLD at the moment.
> >
> > I also appreciate that LLD/ELF development has certainly picked up
pace
> recently and we might actually have a functioning linker sometime soon.
> That is very exciting, and thankyou for all your work.
> >
> > My worry is that the technical debt accrued will have to be paid off
by
> someone at some point. At some point, someone will have to come along and
> make LLD robust ("enough", for some definition), and the current
decision
> will make that person's job much harder.
>
> Again: users *today* are less likely to see a crash in lld than llvm or
> clang and those are much larger code bases.
>
> Cheers,
> Rafael
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://lists.llvm.org/pipermail/llvm-dev/attachments/20160322/f2a827a7/attachment-0001.html>

On 22 March 2016 at 08:26, James Molloy <james at jamesmolloy.co.uk>
wrote:
> Hi,
>
> That's not the point at all. There are known unknowns and unknown
unknowns;
> creating known problems and then not at least marking them as such affects
> people down the line.

Far less than what LLVM has today.

Cheers,
Rafael