search

obuchtala / swig

Branch for the development of SWIG's javascript target (v8 and jsc)
https://github.com/swig/swig
Other
25 stars 14 forks source link

JSCore backend doesn't look C89 clean #29

Open ewmailing opened 10 years ago

ewmailing commented 10 years ago

I've been visually looking at the JSCore output. I don't think it is C89 clean.

For example, I see the use of "bool" as a type which is not standard in C89. And I'm seeing some places where variables are not declared at the top of the scope. (I think SWIG_JSC_NewPointerObj is an example of this.)

This will be a problem for the antiquated Visual Studio C compiler.

obuchtala commented 10 years ago

You are right... I was not careful here. Would you like to fix it?

obuchtala commented 10 years ago

@ewmailing Do u consider this done or should we keep this open?

ewmailing commented 10 years ago

Keep this open for now unless you there was a serious push to clean these all up. What I committed so far was what I saw casually while fixing other problems.

I'm still transitioning to the latest tip. The rebase that happened 6-12 months ago reallly messed us up because we had our own internal multiple branches we were using/fixing things in, and after the rebase, we discovered a lot of regressions so we couldn't merge/switch. (Most of those other fixes I recently submitted were manual transplants we had already fixed in our internal branches a long time ago.)

I'm finally moving things over to the latest (and hoping there will be no rebases), but my initial push is all on Mac/clang (since it has a built-in JSCore), which by default doesn't care about C89.

But a month ago, I was on Windows/Visual Studio, and remember hitting more problems. So I'm anticipating those are still there unless somebody else cleaned them up. I hope to be far enough long to cycle back to Windows/Visual Studio in a couple more weeks.

Also, a quick note, I'm currently not doing any v8 stuff, so I don't know how that looks.