[qcc] qcc post from alt at altamiranus.info requires approval

[Rob Landley]

On 03/24/14 13:54, Alt wrote:
> Please do feel free to elaborate further. I could help out with items
> 1 & 2.A-B, with 2.C being a possibility. 2.D seems a bit complex at
> this stage but I would definitely keep it on my radar.

They pretty much have to be done in sequence. If the project can't build
the simpler packages, it hasn't got a chance of building the more
complex ones.

> I have a full-time job so my email response times may vary, but I am
> willing to commit a few hours per week to this project.
> 
> i also have have access to several VMs that would ease the process of
> building and testing on different platforms.

I've been using http://landley.net/aboriginal for testing.

The first thing you might want to do is read:

http://landley.net/hg/qcc/raw-file/tip/www/tinycc-doc.html

That's the html output of the latex file Fabrice left in his source
tree, describing how his compiler worked.

The first thing _I've_ been meaning to do (and would love if somebody
else did for me) is identify the code in the commits that are _not_
currently able to be BSD licensed, and whether it's easily removed.
(I've been meaning to get that out of the way.)

If it's just documentation or windows support, it's probably easier to
just yank it than try to track down the original submitter.

In this blog entry, I identified the earlier stuff from Fabrice's tenure:

  http://landley.net/notes-2012.html#14-06-2012

The first non-Fabrice commit in the repository is #223 from Peter Lund,
which is here in the repository:

  http://landley.net/hg/qcc/rev/223

We can basically check all of them that way, just change the 223 at the
end to the respective commit.

That one is a patch to the old tcc-doc.texi file, which is probably
worth reading but not necessarily something we have to ship with qcc. In
commit 494 I deleted the tex file and replaced it with the html output,
www/tinycc-doc.html, which I linked to above. (I don't do latex, but am
happy to edit simple html source by hand. So to me, the html _is_ the
preferred format.)

Pater Lund's other commit is comment typo fixes (mostly to old "XXX:
todo" entries that Fabrice later did and removed the comment about. So
those two aren't significant enough to try to track him down.

I was contacted by Daniel Glockener, who asked me to remove his code
(commits 295, 306, 307). That's basically the arm backend that we'd be
replacing with TCC's Tiny Code Generator anyway, so mostly no big loss.

Losing his arm assembler is an issue, but we need a more generic
assembler infrastructure anyway, one capable of producing disassembly
output so the kernel can produce that header file. (I've seen people do
it as shell scripts over the years...)

The rest of the commits in that list, I haven't really looked to see
what they do yet. There are more external commits in the fork I
maintained, but I can identify those once Fabrice's pile is dealt with.
(A lot of it was "the bug report was external, but the code is mine.)

Rob

 1395745121.0