On 9 September 2012 21:43, Jakub Jermar <[email protected]> wrote: > JZr: > - report on his Go port; > - "everything except crypto works"
It was more like "most things should work, but almost nothing is tested, and crypto needs entropy source for its random number generator (parts of crypto that don't depend on entropy are fine)". To expand on that: - Core runtime (which is written in C) has been partly rewritten to suit HelenOS; only single kernel thread is currently supported due to limitations of using fibrils instead of custom scheduler. However this approach means that I can simply reuse the async framework and use libc functions from Go. Some public methods in the "runtime" package are known to be broken, but they are not that important. - The "syscall" and "os" packages have been reimplemented almost from scratch and use libc. "net" is partially rewritten. All the other packages are either completely untouched upstream code or only very slightly patched (most of the code is completely platform independent and thus quite safe to use, provided that the four mentioned packages work correctly). Building included test code is on my TODO list. - Still no clear idea how to eventually integrate Go applications into the HelenOS tree (since the build process is quite different from C), but I think a Go script that generates makefiles is a good solution. That is, by the way, how I build libgo. Async framework binding almost finished, provided I understood the purpose of all the functions correctly. -- Jirka Z. _______________________________________________ HelenOS-devel mailing list [email protected] http://lists.modry.cz/cgi-bin/listinfo/helenos-devel
