2014-11-11 8:39 GMT+02:00 Mark Callow <[email protected]>:

>  juj's objection to library search paths in the end is only about the
> cost of a mistaken search path given as part of the command. I don't think
> it is that big a deal. Following the standard practice of other compilers
> is much more important and makes Emscripten usable with a wider range of
> build tools. The cost can be minimized by either requiring manual download
> of ports (my preference) or asking for confirmation when automatically
> starting a download.
>

I am arguing against the practice that e.g. -lSDL2 would trigger a
download-on-demand if none of the -L paths did not contain a matching file.
That is not standard practice in any compiler environment that I know of.
Like I mentioned before, I am also in favor of the method that downloading
the ports is always manual and requires running a separate install command
first. I don't like going the route of adding interactive confirmation
prompts to the compilation progress, but perhaps we could mitigate the
issue by matching -lXXX to a known port and issuing a hint message that
explains that a library exists in the ports.

I am also arguing against string-matching "-lSDL -> library_sdl.js, -lm ->
/dev/nul, -lal -> library_openal.js, -loal -> library_openal.js, -lsoft_oal
-> library_openal.js, -lGLESv2 -> library_gl.js" and so on, because that
will be error prone and a maintenance mess. I like the idea of reusing
-lxxx.js to mean --js-library library_xxx.js and reusing -L directories to
also search for paths in -lxxx.js, because the explicit suffix .js ensures
there won't be any problems or backwards issues with existing builds
systems.

-- 
You received this message because you are subscribed to the Google Groups 
"emscripten-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to