GCC GSOC 2016

[Manuel López-Ibáñez]

On 01/03/16 19:38, Ayush Goel wrote:
> Hey,

Hi,

Things related to development of GCC are best discussed in gcc@ (not many gcc 
developers actually read gcc-help). I'm moving this discussion here.

> I am interested in contributing to gcc for the gsoc 2016.

Unfortunately, it seems GCC did not apply to participate in GSoC 2016 and the 
deadline passed already: 
https://summerofcode.withgoogle.com/organizations/?sp-search=GCC

It also seems we did not apply last year either (at least 
https://gcc.gnu.org/wiki/SummerOfCode does not show any accepted projects for 
2015).

I think some of us would be interested in mentoring students if they match 
their preferred project [*] (thus, it is better to propose several projects and 
see if a mentor is interested than to try to find a mentor for your preferred 
project).

However, applying to GSoC requires some paperwork and commitment besides 
mentoring, and GCC is lacking developers and existing developers have no free 
time to dedicate to this.

> One of the projects listed a few years back, “Converting different program representations level of GCC back to the source code” seems really interesting to me, and I’d like to discuss the possible ways this could be done. Who should I get into touch with?
>
> IÂ’ve been doing research in extracting call graphs from binaries and analysing them and therefore have gathered sufficient information about Intermediate representations, compiler optimisations. And so feel I might be a good match for the project

My advice to you or any other prospective GSoC student would be:

a) Start publicly working on GCC now: 
https://gcc.gnu.org/wiki/GettingStarted#Basics:_Contributing_to_GCC_in_10_easy_steps

b) Get familiar with GCC devs on your area of interest.

c) Convince them that a project of yours would be so useful and interesting 
that they better spent the time/effort to get GCC in the next GSoC.

d) Once GCC is accepted by GSoC, we get so very few applications that anyone 
with a reasonable project (specially if they already have a willing mentor) is 
almost guaranteed to be accepted.

I understand that the above is not ideal, much less useful for this year, but I 
don't have anything better to offer, sorry. You could also apply to LLVM. They 
are participating in GSoC this year: 
https://summerofcode.withgoogle.com/organizations/?sp-search=LLVM


Good luck,

Manuel.

[*] Projects I would be willing to mentor:

* Replace libiberty with gnulib. See 
http://gcc.gnu.org/ml/gcc-patches/2012-08/msg00362.html
* Anything here: https://gcc.gnu.org/wiki/Better_Diagnostics
* Kill TREE_LIST (https://gcc.gnu.org/wiki/Speedup_areas#Trees)
* Kill TREE_VECTOR
* Kill %qE (not pretty-printing of expressions)
* Kill implicit input_location
* Revive the gdb compile project 
(https://sourceware.org/gdb/wiki/GCCCompileAndExecute), which seems dead.