Re: Generating runtime type information.

[Ori Bernstein <ori@xxxxxxxxxxxxxx>]

On Fri, 27 Mar 2015 14:56:46 -0500
Ryan Gonzalez <rymg19@xxxxxxxxx> wrote:

> That worked. I didn't realize mc used mbld if available.
> 
> Honestly, I would enjoy having mbld *and* libbio in the standard
> distribution.

http://git.eigenstate.org/ori/mc-integrated.git

Take a look, try it out, and make sure that bootstrapping works for you. If
it does, I can push that to the main mc/ repository.
 
> On Fri, Mar 27, 2015 at 2:46 PM, Ori Bernstein <ori@xxxxxxxxxxxxxx> wrote:
> 
> > So, the first thing you're going ot have to rebuild (or delete)
> > is mbld. From there, you should be able to re-bootstrap.
> >
> > I should import mbld and it's dependencies into mc/ for ease of
> > remaking the world.
> >
> > On Fri, 27 Mar 2015 14:24:36 -0500
> > Ryan Gonzalez <rymg19@xxxxxxxxx> wrote:
> >
> > > I can't get it to build though:
> > >
> > > ...
> > > export MYR_MC=../6/6m && \
> > > export MYR_MUSE=../muse/muse && \
> > > ./build.sh -l std alloc.myr bigint.myr bitset.myr blat.myr bytebuf.myr
> > > chartype.myr cmp.myr clear.myr dial.myr die.myr dir.myr dirname.myr
> > > endian.myr env.myr errno.myr execvp.myr extremum.myr fltbits.myr fmt.myr
> > > fltfmt.myr hashfuncs.myr hasprefix.myr hassuffix.myr htab.myr getint.myr
> > > intparse.myr ipparse.myr mk.myr mkpath.myr now.myr option.myr
> > optparse.myr
> > > pathjoin.myr putint.myr rand.myr resolve.myr result.myr search.myr
> > slcp.myr
> > > sldup.myr sleq.myr slfill.myr sljoin.myr slpush.myr slput.myr slurp.myr
> > > sort.myr spork.myr cstrconv.myr strfind.myr strjoin.myr strsplit.myr
> > > strstrip.myr syswrap.myr syswrap-ss.myr swap.myr try.myr types.myr
> > > units.myr utf.myr varargs.myr wait.myr
> > > libsys.a...
> > > libstd.a...
> > > library sys: usefile version unknown
> > > make[1]: *** [libstd.a] Error 1
> > >
> > > I removed sys and libsys.a from /usr/local/lib/myr but still bet the same
> > > error. Also already ran `make clean`.
> > >
> > >
> > > On Fri, Mar 27, 2015 at 1:19 AM, Ori Bernstein <ori@xxxxxxxxxxxxxx>
> > wrote:
> > >
> > > > Congratulations everyone! you get to rebuild as we break ABI
> > > > again!
> > > >
> > > > We now generate type descriptions for variadic types, and push
> > > > a pointer to the variadic arguments on to the stack when we call.
> > > > The valist code currently ignores this, and does nothing useful
> > > > with it -- this should be fixed.
> > > >
> > > > In other words, if you have a variadic function f:
> > > >
> > > >         const f : (a : int, b : byte[:], ... -> void)
> > > >
> > > > and you call it like this:
> > > >
> > > >         f(1, "asdf", 'c', "blah", false)
> > > >
> > > > you will get, on the stack:
> > > >
> > > >         [
> > > >         1       : int,
> > > >         "asdf"  : byte[:],
> > > >         argtype : byte#
> > > >         'c'     : char,
> > > >         "blah"  : byte[:],
> > > >         false   : bool
> > > >         ]
> > > >
> > > > The argtype will be a binary encoded description of the types
> > > > of the tuple (char, byte[:], bool)
> > > >
> > > > This means that you can write (admittedly, slightly hairy) code
> > > > to parse out the types passed in for variadics, and write code
> > > > that looks like:
> > > >
> > > >         const pack : (vals : ... -> byte[:])
> > > >
> > > > One goal that I would like to support is making std.fmt()
> > > > callable as such:
> > > >
> > > >         std.fmt("%: list contains %\n", 1, [1,2,3][:])
> > > >
> > > >
> > > > --
> > > > Ori Bernstein <ori@xxxxxxxxxxxxxx>
> > > >
> > > >
> > >
> > >
> > > --
> > > Ryan
> > > [ERROR]: Your autotools build scripts are 200 lines longer than your
> > > program. Something’s wrong.
> > > http://kirbyfan64.github.io/
> >
> >
> > --
> > Ori Bernstein <ori@xxxxxxxxxxxxxx>
> >
> 
> 
> 
> -- 
> Ryan
> [ERROR]: Your autotools build scripts are 200 lines longer than your
> program. Something’s wrong.
> http://kirbyfan64.github.io/


-- 
Ori Bernstein <ori@xxxxxxxxxxxxxx>