[GSOC] Self-introduction

[Philippe Ombredanne]

On Mon, Mar 21, 2016 at 12:07 AM, Grace O'Hair-Sherman
<gohairsh at ucsc.edu> wrote:
> Hello!
>
> I’m a Computer Science major and a member of my university’s Linux
> Users’ Group and I’m interested in contributing to strace,
> specifically the Structured output project involving parsing strace
> output as JSON as suggested by Philippe Ombredanne.
>
> I’ve used and scripted in BASH for almost two years, and I am
> proficient in Git and JSON. I have experience writing JSON by hand and
> writing a Python script to automatically generate it from comma
> separated values. I have a working knowledge of C. Although I lack a
> lot of experience with it, I’m a fast and thorough learner and I will
> put in the time to improve my skills.
>
> Right now I’m preparing by looking at your repository and brushing up
> on C. Any suggestions for other preparation I can do?
>
> I’m really excited for the opportunity to contribute to such a
> long-running open source project! Thank you for your time.
>
> Sincerely,
> Grace O’Hair-Sherman
>
> gohairsh at ucsc.edu
> University of California, Santa Cruz
> Baskin School of Engineering
> Computer Science

Hello and welcome to strace!
Thank you for considering our little project for your GSOC application.

As I said in recent posts, producing a structured output would not be
for the faint of heart.
Not so much in terms of C code complexity but because of the magnitude
of changes needed:
these would eventually touch a very large portion of the code and the
difficulty is:
- to be able complete something useful and meaningful within the GSOC timeframe,
- for us to review and accept a larger volume of patches, and
- to ensure that we reach a steady state where the patches form a
consistent whole rather than a half-baked solution.

That said and for a start, you should check out previous post on the
topic in particular the threads and patches and links from the GSOC
2014.
And of course try to build strace.

Now there is also an alternative project we could consider:
Instead of spitting JSON from strace directly, you could build a
parser of already generated traces to spit JSON. While this could
likely never be as good as a proper structured output, it good go a
long way to help many. And this would be something to code in Python
(or possibly Perl?, or...) rather than C.
It could be much easier to swallow as a GSOC project, and you seem to
have some Python skills...

-- 
Cordially
Philippe Ombredanne