Contribution to open source projects...

[The_GTA]

I have had experience with three open source projects where I created pull-requests to. My conclusion is that doing so is a big hassle.

 

Example #1: AsmJIT

​https://github.com/asmjit/asmjit/pull/173

 

Purpose:

I created an addon to the AsmJIT assembly generation framework which gave feedback about memory-location-sensity assembly output so that it could be fixed up. With this power I can create /DYNAMICBASE style PE executables from assembly.

 

My experience:

The author of AsmJIT entangled me into a discussion where he wanted to work on an text assembler. He claimed that my work had to be compatible with his vision before it were merged. Thus my pull is forever pending.

 

Example #2: Compressonator

https://github.com/GPUOpen-Tools/Compressonator/pull/23

 

Purpose:

Added support for Linux operating system.

 

My experience:

The commercial guys at AMD, one of which is Navin Patel, have been pretty busy with things. Unfortunately they have never finished evaluating my pull request which is weird. I would assume that working code should be merged and one could easily work based on it. Since I am a student who just did it for fun I am stressed but will look how to get this up-to-date.

 

Example #3: SAO

​​pull request is ripperoni in pepperoni

 

Purpose:

Contribution of my PE executable parsing framework (PEframework) for SAO game loading so that everything was nicely documented.

 

My experience:

iFarbod was not convinced of the necessity of my code so he declined it. I can fully understand that he wanted to do it himself to learn so I let it pass by. Not really sure if he had any code quality objections for my work. I was sort of disappointed. But okay.

 

------------------------

 

Conclusion: think twice before you get into this hassle.

[K^2]

Yeah, open source contribs often go that way. We're using a few open source libraries at work. At least two of my co-workers have tried getting bug fixes we had to make pulled into main branch with variable success. But it's always going to be some amount of hassle and red-tape on a project made up of more than one person.

 

Personally, I've ran into something very similar while working at Google with the internal code. Any fixes I had to put into other team's libraries would go through more scrutiny than they often deserved. My favorite example is when I've spotted that leap seconds counter wasn't updated for the most recent leap second in one of the libraries I was considering to use. Scope of the project: Increment integer by one. Time to get approved: Two weeks and three revisions.