[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Tinycc-devel] modern c++ compiler written in C

From: u-tcc-uepj
Subject: Re: [Tinycc-devel] modern c++ compiler written in C
Date: Thu, 14 May 2015 09:16:59 +0200

Hello Sergey,

On Thu, May 14, 2015 at 04:17:17AM +0300, Sergey Korshunoff wrote:
> 2015-04-29 9:55 GMT+03:00, address@hidden
> > I worked around most of the omissions in the only available archive
> > of the latest cfront source and modified it to be testable/usable with
> > current C compilers and programs (namely taught it to grok long long),
> > it works as it should.
> Can you share your work with me? I failed to get a hello_word compiled by it
> and switched to v1.0 with patches by grishka.  With this I can compile
> a hello_world.

The cfront 3.0.3 archive at http://www.softwarepreservation.org/ is
unfortunately partly broken and incomplete, despite being assumed to be
the official distribution media. I guess nobody verified the media at
the time and many of the customers either already had the missing pieces
from earlier versions or assumed they did something wrong themselves
and gave up on cfront.

Nevertheless with a number of fixes it is possible to bootstrap the
compiler and also build some of the libraries.

An additional and important part is to make cfront compatible with
the available C compilers, otherwise you'd have to reconstruct the whole
ancient environment, not only the part which you are to research.

This is what the attached patch is for.

I may also try to collect and post the other fixes which were necessary
to bootstrap cfront, I did this with (modern) pcc and musl as the libc.

A toolchain of pcc and cfront is both compact and impressively fast.
With tcc this would be quite certainly with a very large margin the
fastest c++ compiler ever. :)

As for the language dialect, the projects I care about do not rely on
the features added in the later c++ versions.

May be, if the interest for revived cfront becomes noticeable, AT&T might
become interested in some extra goodwill and would take the effort needed
to make it officially free?

They can certainly not gain anything by forbidding its use, it is just
the decision making for the license change which is hard/expensive.


Attachment: cfront.long_long.patch.2014-09-12.bz2
Description: Binary data

reply via email to

[Prev in Thread] Current Thread [Next in Thread]