Re: upgrading GCC 4.1 branch

>>> The major problem is C++ frontends.
>> What do we have that needs C++?
> In base?  groff is the biggest issue, because it's big
and messy and
> sane people don't want to go near it.

  Is
there a publicly available spec for the *roff language? 
I'd
like to have a look and at least consider building an
implementation in
C.  (When I find my code in tones of trouble, / Friends and
colleagues
come to me, / Speaking words of wisdom:  "Write in
C".)

> But realistically there are enough C++ things in pkgsrc
that if
> there's no C++ compiler in base, there will always be
one installed
> in /usr/pkg, which doesn't seem like much of an
improvement.

It does to me.  You could say similar things about perl, or
GNU make,
or various other things - but they're not in base and I
don't think
they should be.  (Besides, except for manpage formatting, I
don't think
*I* use anything requiring C++... 

> Convincing the world to migrate onward from C++ is an
option too, in
> the long run, but that requires first inventing a
credible migration
> path.

Moving our C++ compiler to pkgsrc might provide a little
incentive....

/~ The ASCII				der Mouse
 / Ribbon Campaign
 X  Against HTML	       mouserodents.montreal.qc.ca
/  Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3
27 4B

On Tue, Feb 05, 2008 at 02:00:55PM -0500, der Mouse wrote:
> >>> The major problem is C++ frontends.
> >> What do we have that needs C++?
> > In base?  groff is the biggest issue, because it's
big and messy and
> > sane people don't want to go near it.
> 
>  
Is there a publicly available spec for the *roff language? 
I'd
> like to have a look and at least consider building an
implementation in
> C.  (When I find my code in tones of trouble, / Friends
and colleagues
> come to me, / Speaking words of wisdom:  "Write in
C".)

ditroff is available under CDDL. The older roff from 32V is
under BSD
license. The former needs around a third of the time for
gcc(1).

> > Convincing the world to migrate onward from C++ is
an option too, in
> > the long run, but that requires first inventing a
credible migration
> > path.
> 
> Moving our C++ compiler to pkgsrc might provide a
little incentive....

It moves a hard to to maintain piece to a place where even
less hands
are present. Sounds like a good idea.

Joerg

On Tue, Feb 05, 2008 at 02:00:55PM -0500, der Mouse wrote:
 > >> What do we have that needs C++?
 > > In base?  groff is the biggest issue, because
it's big and messy and
 > > sane people don't want to go near it.
 > 
 > 
 Is there a publicly available spec for the *roff language? 
I'd
 > like to have a look and at least consider building an
implementation in
 > C.  (When I find my code in tones of trouble, /
Friends and colleagues
 > come to me, / Speaking words of wisdom:  "Write
in C".)

You're mad 

Also, if you're going to go to the trouble of writing a new
typesetting engine, can you please please please come up
with a sane
frontend? I would much rather hand-convert every man page in
base
using ed than breathe new life into roff.

 > > But realistically there are enough C++ things in
pkgsrc that if
 > > there's no C++ compiler in base, there will
always be one installed
 > > in /usr/pkg, which doesn't seem like much of an
improvement.
 > 
 > It does to me.

Not if the objection is to license toxicity.

 > > Convincing the world to migrate onward from C++
is an option too, in
 > > the long run, but that requires first inventing a
credible migration
 > > path.
 > 
 > Moving our C++ compiler to pkgsrc might provide a
little incentive....

To do what? Design a suitable new language and write a
compiler for
it, then try to convince the world it's better than C++?
Probably not.

-- 
David A. Holland
dhollandnetbsd.org