GCC museum?

GCC museum?

[Andrew Morton]

Folks,

Back in '94 I ported gcc/g++ 2.5.8 to the TMS34010.  This was a
specialised CISC graphics processor from Texas Instruments.

The 34k was bit-addressable.  Perhaps this is still the only port of GCC
to a bit-addressable architecture?

It was a _lot_ of work - in many places the compiler assumes that adding
one to a pointer advances it by eight bits, not by one bit.

Apart from doing the md and tm.h stuff, most of the work was in hunting
down and nailing those assumptions.  This would still be useful for
someone who was porting a current version to a bit-addressable
architecture.

So.  Is anyone interested in taking this historical curiosity off my
hands, or do I take it to my grave?


Cheers,
Andrew.

Re: GCC museum?

[Nick Ing-Simmons]

Andrew Morton <morton@nortelnetworks.com> writes:
>Folks,
>
>Back in '94 I ported gcc/g++ 2.5.8 to the TMS34010.  This was a
>specialised CISC graphics processor from Texas Instruments.
>
>The 34k was bit-addressable.  Perhaps this is still the only port of GCC
>to a bit-addressable architecture?

If it worked it is the maybe the only _working_ port.

>
>It was a _lot_ of work - in many places the compiler assumes that adding
>one to a pointer advances it by eight bits, not by one bit.

Which is why mine never got finished :-(

>
>Apart from doing the md and tm.h stuff, most of the work was in hunting
>down and nailing those assumptions.  This would still be useful for
>someone who was porting a current version to a bit-addressable
>architecture.

Or a word-addressed architecture. The UNIT == BYTE stuff bit on the 
32-bit-words-only DSPs too.

>
>So.  Is anyone interested in taking this historical curiosity off my
>hands, or do I take it to my grave?
>
>
>Cheers,
>Andrew.
-- 
Nick Ing-Simmons

Re: GCC museum?

[Marty Leisner]

Andrew,

In about 1989 I tried to do a port with gcc 1.x to the TI34010 -- I found the 
TI C compiler amazingly crude (poor?).

I gave up after I saw how hard it would be (after I started generating a 
reasonable amount
of code to teach me how to write a  machine description...)

Does the current gcc have hooks in their for bit aligned
processors?

Are there any bit aligned processors currently of interest?

I think you should put the work in a public, safe place, in case anyone
needs it...


Marty Leisner


"Andrew Morton" <morton@nortelnetworks.com> writes  on Sun, 17 Oct 1999 12:11:4
4 -0000
     > Folks,
     > 
     > Back in '94 I ported gcc/g++ 2.5.8 to the TMS34010.  This was a
     > specialised CISC graphics processor from Texas Instruments.
     > 
     > The 34k was bit-addressable.  Perhaps this is still the only port of GCC
     > to a bit-addressable architecture?
     > 
     > It was a _lot_ of work - in many places the compiler assumes that adding
     > one to a pointer advances it by eight bits, not by one bit.
     > 
     > Apart from doing the md and tm.h stuff, most of the work was in hunting
     > down and nailing those assumptions.  This would still be useful for
     > someone who was porting a current version to a bit-addressable
     > architecture.
     > 
     > So.  Is anyone interested in taking this historical curiosity off my
     > hands, or do I take it to my grave?
     > 
     > 
     > Cheers,
     > Andrew.

Re: GCC museum?

[Jeffrey A Law]

  In message < 199910180426.AAA02658@rochester.rr.com >you write:
  > Does the current gcc have hooks in their for bit aligned
  > processors?
Not that I'm aware of.

  > I think you should put the work in a public, safe place, in case anyone
  > needs it...
If someone puts it up, make sure we get a URL so that we can put a link to
it on the web page.  There's a lot of weird ports and and few languages that
are not integrated that we can/should at keep a pointer to their source on
the web site.

jeff

Re: GCC museum?

[Andrew Morton]

Marty Leisner wrote:
> 
> Andrew,
> 
> In about 1989 I tried to do a port with gcc 1.x to the TI34010 -- I found the
> TI C compiler amazingly crude (poor?).

Plus it ran under Windows..

> I gave up after I saw how hard it would be (after I started generating a
> reasonable amount
> of code to teach me how to write a  machine description...)
> 
> Does the current gcc have hooks in their for bit aligned
> processors?

As Jeffrey has pointed out, it doesn't.  I recall that it loses info at
quite a high level.  For example,

	char *p; int i;
	p[i];

is recognised and expanded in cp-typeck.c (now cp/typeck.c) and the info
which the backend needed to multiply the subscript by 8 was sort of lost
(lost to me, anyway).  

> Are there any bit aligned processors currently of interest?

Not that I'm aware of.  The TMS34010 is "obsolete" and the '020 is "not
for new designs".

http://www.ti.com/sc/docs/products/process/tms34010.html
http://www.ti.com/sc/docs/products/process/tms34020a.html

> I think you should put the work in a public, safe place, in case anyone
> needs it...

OK, I'll do a little web page.  I'll put up all the others necessities
of life also: as34, ld34, nm34, dis34, size34, strip34, pc34. (Hey, it
all still compiles!)

Re: GCC museum?

[Andrew Morton]

Nick Ing-Simmons wrote:
> 
> Andrew Morton <morton@nortelnetworks.com> writes:
> >Folks,
> >
> >Back in '94 I ported gcc/g++ 2.5.8 to the TMS34010.  This was a
> >specialised CISC graphics processor from Texas Instruments.
> >
> >The 34k was bit-addressable.  Perhaps this is still the only port of GCC
> >to a bit-addressable architecture?
> 
> If it worked it is the maybe the only _working_ port.

Yes, it worked.  Well enough to develop a 20kloc C++ embedded system:
RTOS, graphics, etc.  A variant of a video poker machine.  I barely used
the C compiler, though.

Re: GCC museum?

[Andrew Morton]

I wrote:
> 
> OK, I'll do a little web page.

http://www.uow.edu.au/~andrewm/gsptools.html

This is a stable URL.

Re: GCC museum?

[Nick Ing-Simmons]

Andrew Morton <morton@nortelnetworks.com> writes:
>Folks,
>
>Back in '94 I ported gcc/g++ 2.5.8 to the TMS34010.  This was a
>specialised CISC graphics processor from Texas Instruments.
>
>The 34k was bit-addressable.  Perhaps this is still the only port of GCC
>to a bit-addressable architecture?

If it worked it is the maybe the only _working_ port.

>
>It was a _lot_ of work - in many places the compiler assumes that adding
>one to a pointer advances it by eight bits, not by one bit.

Which is why mine never got finished :-(

>
>Apart from doing the md and tm.h stuff, most of the work was in hunting
>down and nailing those assumptions.  This would still be useful for
>someone who was porting a current version to a bit-addressable
>architecture.

Or a word-addressed architecture. The UNIT == BYTE stuff bit on the 
32-bit-words-only DSPs too.

>
>So.  Is anyone interested in taking this historical curiosity off my
>hands, or do I take it to my grave?
>
>
>Cheers,
>Andrew.
-- 
Nick Ing-Simmons

Re: GCC museum?

[Jeffrey A Law]

  In message < 199910180426.AAA02658@rochester.rr.com >you write:
  > Does the current gcc have hooks in their for bit aligned
  > processors?
Not that I'm aware of.

  > I think you should put the work in a public, safe place, in case anyone
  > needs it...
If someone puts it up, make sure we get a URL so that we can put a link to
it on the web page.  There's a lot of weird ports and and few languages that
are not integrated that we can/should at keep a pointer to their source on
the web site.

jeff

Re: GCC museum?

[Andrew Morton]

I wrote:
> 
> OK, I'll do a little web page.

http://www.uow.edu.au/~andrewm/gsptools.html

This is a stable URL.

Re: GCC museum?

[Andrew Morton]

Nick Ing-Simmons wrote:
> 
> Andrew Morton <morton@nortelnetworks.com> writes:
> >Folks,
> >
> >Back in '94 I ported gcc/g++ 2.5.8 to the TMS34010.  This was a
> >specialised CISC graphics processor from Texas Instruments.
> >
> >The 34k was bit-addressable.  Perhaps this is still the only port of GCC
> >to a bit-addressable architecture?
> 
> If it worked it is the maybe the only _working_ port.

Yes, it worked.  Well enough to develop a 20kloc C++ embedded system:
RTOS, graphics, etc.  A variant of a video poker machine.  I barely used
the C compiler, though.

GCC museum?

[Andrew Morton]

Folks,

Back in '94 I ported gcc/g++ 2.5.8 to the TMS34010.  This was a
specialised CISC graphics processor from Texas Instruments.

The 34k was bit-addressable.  Perhaps this is still the only port of GCC
to a bit-addressable architecture?

It was a _lot_ of work - in many places the compiler assumes that adding
one to a pointer advances it by eight bits, not by one bit.

Apart from doing the md and tm.h stuff, most of the work was in hunting
down and nailing those assumptions.  This would still be useful for
someone who was porting a current version to a bit-addressable
architecture.

So.  Is anyone interested in taking this historical curiosity off my
hands, or do I take it to my grave?


Cheers,
Andrew.

Re: GCC museum?

[Marty Leisner]

Andrew,

In about 1989 I tried to do a port with gcc 1.x to the TI34010 -- I found the 
TI C compiler amazingly crude (poor?).

I gave up after I saw how hard it would be (after I started generating a 
reasonable amount
of code to teach me how to write a  machine description...)

Does the current gcc have hooks in their for bit aligned
processors?

Are there any bit aligned processors currently of interest?

I think you should put the work in a public, safe place, in case anyone
needs it...


Marty Leisner


"Andrew Morton" <morton@nortelnetworks.com> writes  on Sun, 17 Oct 1999 12:11:4
4 -0000
     > Folks,
     > 
     > Back in '94 I ported gcc/g++ 2.5.8 to the TMS34010.  This was a
     > specialised CISC graphics processor from Texas Instruments.
     > 
     > The 34k was bit-addressable.  Perhaps this is still the only port of GCC
     > to a bit-addressable architecture?
     > 
     > It was a _lot_ of work - in many places the compiler assumes that adding
     > one to a pointer advances it by eight bits, not by one bit.
     > 
     > Apart from doing the md and tm.h stuff, most of the work was in hunting
     > down and nailing those assumptions.  This would still be useful for
     > someone who was porting a current version to a bit-addressable
     > architecture.
     > 
     > So.  Is anyone interested in taking this historical curiosity off my
     > hands, or do I take it to my grave?
     > 
     > 
     > Cheers,
     > Andrew.

Re: GCC museum?

[Andrew Morton]

Marty Leisner wrote:
> 
> Andrew,
> 
> In about 1989 I tried to do a port with gcc 1.x to the TI34010 -- I found the
> TI C compiler amazingly crude (poor?).

Plus it ran under Windows..

> I gave up after I saw how hard it would be (after I started generating a
> reasonable amount
> of code to teach me how to write a  machine description...)
> 
> Does the current gcc have hooks in their for bit aligned
> processors?

As Jeffrey has pointed out, it doesn't.  I recall that it loses info at
quite a high level.  For example,

	char *p; int i;
	p[i];

is recognised and expanded in cp-typeck.c (now cp/typeck.c) and the info
which the backend needed to multiply the subscript by 8 was sort of lost
(lost to me, anyway).  

> Are there any bit aligned processors currently of interest?

Not that I'm aware of.  The TMS34010 is "obsolete" and the '020 is "not
for new designs".

http://www.ti.com/sc/docs/products/process/tms34010.html
http://www.ti.com/sc/docs/products/process/tms34020a.html

> I think you should put the work in a public, safe place, in case anyone
> needs it...

OK, I'll do a little web page.  I'll put up all the others necessities
of life also: as34, ld34, nm34, dis34, size34, strip34, pc34. (Hey, it
all still compiles!)

Re: GCC museum?

[Gerald Pfeifer]

On Mon, 18 Oct 1999, Andrew Morton wrote:
> http://www.uow.edu.au/~andrewm/gsptools.html
> 
> This is a stable URL.

Hi Andrew, I have now crafted an entry (as well as a new category) for our
extensions.html page. If you have any suggestions on how to improve that,
just let me know!

Thanks for letting us know about this port and providing a nice page!

Gerald

PS: And sorry for the delay, but -- finally -- my GCC web pages TODO list
is getting smaller and smaller. ;-)


Index: extensions.html
===================================================================
RCS file: /cvs/gcc/wwwdocs/htdocs/extensions.html,v
retrieving revision 1.2
diff -c -3 -p -r1.2 extensions.html
*** extensions.html	1999/11/10 00:45:23	1.2
--- extensions.html	1999/12/17 20:36:46
*************** installation on non-Sun systems or when 
*** 48,57 ****
  <p>Contact: <a href=" mailto:hp@axis.se ">hp@axis.se</a></p>
  
  <hr>
  <p><a href="index.html">Return to the GCC home page</a>
  
  <hr>
! <p><i>Last modified:  November 9, 1999</i>
  
  </body>
  </html>
--- 48,71 ----
  <p>Contact: <a href=" mailto:hp@axis.se ">hp@axis.se</a></p>
  
  <hr>
+ 
+ <h1>GCC Museum</h1>
+ 
+ <p>In this section you will find extensions which do not (directly)
+ apply to current GCC sources but are of historical interest or may
+ be helpful to consult for future development.</p>
+ 
+ <h2><a href=" http://www.uow.edu.au/~andrewm/gsptools.html ">Development
+ tools for the TMS340x0</a></h2>
+ 
+ <p>A GCC port to a <EM>bit-addressable</EM> architecture.</p>
+ 
+ 
+ <hr>
  <p><a href="index.html">Return to the GCC home page</a>
  
  <hr>
! <p><i>Last modified:  December 17, 1999</i>
  
  </body>
  </html>

Re: GCC museum?

[Rich Churcher]

Gerald Pfeifer writes:
 > 
 > PS: And sorry for the delay, but -- finally -- my GCC web pages TODO list
 > is getting smaller and smaller. ;-)

If you need any help whittling it down, please let me know :o)

Cheers,

-- 
Rich
churchmouse@mpx.com.au

Re: GCC museum?

[Gerald Pfeifer]

On Sun, 19 Dec 1999, Rich Churcher wrote:
> > PS: And sorry for the delay, but -- finally -- my GCC web pages TODO list
> > is getting smaller and smaller. ;-)
> If you need any help whittling it down, please let me know :o)

Well, you asked for it. :-)

It would be great if you could have a close look at those pages linked to
under the "Development" heading of our main page, especially contribute.html,
cvs.html, cvswrite.html, and checkin.html.

 . Can we compress the information there a bit?

 . Should we add further information?

 . Do we need a couple of further cross-links between these pages?

 . Should we unify some of these pages? For example, cvs.html and
   cvswrite.html?

 . Is the information there clear for a non-expert?
   Is it easy enough for an expert to find relevant information quickly?

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/

Re: GCC museum?

[Rich Churcher]

Gerald Pfeifer writes:
 > On Sun, 19 Dec 1999, Rich Churcher wrote:
 > > > PS: And sorry for the delay, but -- finally -- my GCC web pages TODO list
 > > > is getting smaller and smaller. ;-)
 > > If you need any help whittling it down, please let me know :o)
 > 
 > Well, you asked for it. :-)

That's ok, I meant it :o) I enjoyed helping out the last time (with
the FAQ), but due to moving countries and jobs I never got around to
writing back and asking for more "assignments".  I use GCC every day,
and always feel like I should be contributing more to the project.


 > It would be great if you could have a close look at those pages linked to
 > under the "Development" heading of our main page, especially contribute.html,
 > cvs.html, cvswrite.html, and checkin.html.
 > 
 >  . Can we compress the information there a bit?
 > 
 >  . Should we add further information?
 > 
 >  . Do we need a couple of further cross-links between these pages?
 > 
 >  . Should we unify some of these pages? For example, cvs.html and
 >    cvswrite.html?
 > 
 >  . Is the information there clear for a non-expert?
 >    Is it easy enough for an expert to find relevant information quickly?

I'll have a look and let you know.

-- 
Rich
churchmouse@mpx.com.au

Re: GCC museum?

[Rich Churcher]

Gerald Pfeifer writes:
 > 
 > It would be great if you could have a close look at those pages linked to
 > under the "Development" heading of our main page, especially contribute.html,
 > cvs.html, cvswrite.html, and checkin.html.
 > 
 >  . Can we compress the information there a bit?

I've had a chance to look over these pages.  Actually, everything
seems solid and well-organised, so any changes will be incremental.

 > 
 >  . Should we add further information?

This is perhaps not the best question for me to answer.  I qualify for 
the "non-expert" point of view, as referenced below :o)  However, I
promise faithfully to add anything others consider relevant into any
patch I submit.  It looks complete to my untrained eyes.

 
 >  . Do we need a couple of further cross-links between these pages?

Maybe.  A link to cvs.html under the Submitting Patches heading in
contribute.html wouldn't hurt.  If you're going to merge the cvs
pages, you don't need much else.


 >  . Should we unify some of these pages? For example, cvs.html and
 >    cvswrite.html?

I think it would be a good idea to merge cvswrite.html and
contribute.html, which cover similar territory.  cvs.html is sort of
different, in that you could have read access purely as a matter of
interest, without needing to know about write access and procedures.


 >  . Is the information there clear for a non-expert?

That's me :o)  I find it easy enough to follow.  One thing I did find
a wee bit excessive was the description of a cvs check-in under
"Example check-in session" in checkin.html.  While this was very useful, I 
suggest providing a separate page and linking to it.  But then, my
preference for web pages is bite-sized pieces rather than lots of
scrolling - YMMV.


 >    Is it easy enough for an expert to find relevant information
 > quickly?

Again, how would I know?  Ok, I'm being obtuse.  Let me put it another 
way - I suspect that experts will already know the logistics, and just 
need a few details.  If you like, I can put those at the top and
further explanations further down on each page.  Is that the kind of
structure that would make sense to a community like GCC?  Or a link at 
the top of the page marked "Essential details", or some such.

A general comment on presentation: I think it would make sense for
these pages to move towards white backgrounds, black text.  Some are,
some aren't, and it looks wierd.  I presume some may have been
automatically generated from .texi, or whatever.  Still - compare
projects.html with cvs.html, for example.

Finally - should I be discussing this on gcc-patches?  Sorry, it's
been awhile...

Cheers,

-- 
Rich
churchmouse@mpx.com.au

Web pages (was: GCC museum?)

[Gerald Pfeifer]

On Thu, 30 Dec 1999, Rich Churcher wrote:
> A link to cvs.html under the Submitting Patches heading in
> contribute.html wouldn't hurt.

Hmm, there is already a link to "CVS repository" there, do you think
we should make the link more prominent?

(Some people still use snapshots and submit patches based one these.)

>>  . Should we unify some of these pages? For example, cvs.html and
>>    cvswrite.html?
> I think it would be a good idea to merge cvswrite.html and
> contribute.html, which cover similar territory.

Sounds like a good idea to me!

But you did mean to write cvswrite.html and checkin.html, didn't you?

> cvs.html is sort of different, in that you could have read access
> purely as a matter of interest, without needing to know about write
> access and procedures.

Good point!

> That's me :o)  I find it easy enough to follow.  One thing I did find
> a wee bit excessive was the description of a cvs check-in under
> "Example check-in session" in checkin.html.  While this was very
> useful, I suggest providing a separate page and linking to it.  But
> then, my preference for web pages is bite-sized pieces [...]

Perhaps we can keep it at the very end of the unified
cvswrite+contribute.html page and add a small index at the beginning
of the page, something like the following?

 <A HREF="heading1">Heading 1</A> | <A HREF="heading2">Heading 2</A>...

>> Is it easy enough for an expert to find relevant information quickly?
> Again, how would I know?  Ok, I'm being obtuse.  Let me put it another 
> way - I suspect that experts will already know the logistics, and just 
> need a few details.  If you like, I can put those at the top and
> further explanations further down on each page.  Is that the kind of
> structure that would make sense to a community like GCC?  Or a link at 
> the top of the page marked "Essential details", or some such.

First the general stuff and then the those details would be my preference.

Essentially the "Setting up your local CVS tree" section of cvswrite.html
already contains the important information for experts.

> A general comment on presentation: I think it would make sense for
> these pages to move towards white backgrounds, black text.  Some are,
> some aren't, and it looks wierd.

Yes, yes, and yes. Indeed that's what I am currently working on. :-)
  http://gcc.gnu.org/ml/gcc-patches/1999-12/msg00968.html

Let's make a deal: I take care of this issue, and you attack our cvs
web pages. ;-)

> Finally - should I be discussing this on gcc-patches?  Sorry, it's
> been awhile...

Before there's a patch, this list is apropriate, AFAIS. But I'm looking
forward to receiving a patch via gcc-patches! :-) 

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/

Re: Web pages (was: GCC museum?)

[Rich Churcher]

On Thu, 30 Dec 1999, you wrote:
> On Thu, 30 Dec 1999, Rich Churcher wrote:
> > A link to cvs.html under the Submitting Patches heading in
> > contribute.html wouldn't hurt.
> 
> Hmm, there is already a link to "CVS repository" there, do you think
> we should make the link more prominent?

Whups, so there is... I thought that went somewhere else.  Already you
can see how reliable and error-free I am ;o)   Perhaps pointing this to
cvswrite/checkin.html might be more useful.


> >>  . Should we unify some of these pages? For example, cvs.html and
> >>    cvswrite.html?
> > I think it would be a good idea to merge cvswrite.html and
> > contribute.html, which cover similar territory.
> 
> Sounds like a good idea to me!
> 
> But you did mean to write cvswrite.html and checkin.html, didn't you?

Yes *sigh*  :o)

 
> > That's me :o)  I find it easy enough to follow.  One thing I did find
> > a wee bit excessive was the description of a cvs check-in under
> > "Example check-in session" in checkin.html.  While this was very
> > useful, I suggest providing a separate page and linking to it.  But
> > then, my preference for web pages is bite-sized pieces [...]
> 
> Perhaps we can keep it at the very end of the unified
> cvswrite+contribute.html page and add a small index at the beginning
> of the page, something like the following?
> 
>  <A HREF="heading1">Heading 1</A> | <A HREF="heading2">Heading 2</A>...

Sure, I'll take care of that.

 
> > A general comment on presentation: I think it would make sense for
> > these pages to move towards white backgrounds, black text.  Some are,
> > some aren't, and it looks wierd.
> 
> Yes, yes, and yes. Indeed that's what I am currently working on. :-)
>   http://gcc.gnu.org/ml/gcc-patches/1999-12/msg00968.html
> 
> Let's make a deal: I take care of this issue, and you attack our cvs
> web pages. ;-)

:o)  It's a deal.

Cheers,

--
Rich
churchmouse@mpx.com.au

Re: GCC museum?

[Gerald Pfeifer]

On Mon, 18 Oct 1999, Andrew Morton wrote:
> http://www.uow.edu.au/~andrewm/gsptools.html
> 
> This is a stable URL.

Hi Andrew, I have now crafted an entry (as well as a new category) for our
extensions.html page. If you have any suggestions on how to improve that,
just let me know!

Thanks for letting us know about this port and providing a nice page!

Gerald

PS: And sorry for the delay, but -- finally -- my GCC web pages TODO list
is getting smaller and smaller. ;-)


Index: extensions.html
===================================================================
RCS file: /cvs/gcc/wwwdocs/htdocs/extensions.html,v
retrieving revision 1.2
diff -c -3 -p -r1.2 extensions.html
*** extensions.html	1999/11/10 00:45:23	1.2
--- extensions.html	1999/12/17 20:36:46
*************** installation on non-Sun systems or when 
*** 48,57 ****
  <p>Contact: <a href=" mailto:hp@axis.se ">hp@axis.se</a></p>
  
  <hr>
  <p><a href="index.html">Return to the GCC home page</a>
  
  <hr>
! <p><i>Last modified:  November 9, 1999</i>
  
  </body>
  </html>
--- 48,71 ----
  <p>Contact: <a href=" mailto:hp@axis.se ">hp@axis.se</a></p>
  
  <hr>
+ 
+ <h1>GCC Museum</h1>
+ 
+ <p>In this section you will find extensions which do not (directly)
+ apply to current GCC sources but are of historical interest or may
+ be helpful to consult for future development.</p>
+ 
+ <h2><a href=" http://www.uow.edu.au/~andrewm/gsptools.html ">Development
+ tools for the TMS340x0</a></h2>
+ 
+ <p>A GCC port to a <EM>bit-addressable</EM> architecture.</p>
+ 
+ 
+ <hr>
  <p><a href="index.html">Return to the GCC home page</a>
  
  <hr>
! <p><i>Last modified:  December 17, 1999</i>
  
  </body>
  </html>

Re: GCC museum?

[Gerald Pfeifer]

On Sun, 19 Dec 1999, Rich Churcher wrote:
> > PS: And sorry for the delay, but -- finally -- my GCC web pages TODO list
> > is getting smaller and smaller. ;-)
> If you need any help whittling it down, please let me know :o)

Well, you asked for it. :-)

It would be great if you could have a close look at those pages linked to
under the "Development" heading of our main page, especially contribute.html,
cvs.html, cvswrite.html, and checkin.html.

 . Can we compress the information there a bit?

 . Should we add further information?

 . Do we need a couple of further cross-links between these pages?

 . Should we unify some of these pages? For example, cvs.html and
   cvswrite.html?

 . Is the information there clear for a non-expert?
   Is it easy enough for an expert to find relevant information quickly?

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/

Re: Web pages (was: GCC museum?)

[Rich Churcher]

On Thu, 30 Dec 1999, you wrote:
> On Thu, 30 Dec 1999, Rich Churcher wrote:
> > A link to cvs.html under the Submitting Patches heading in
> > contribute.html wouldn't hurt.
> 
> Hmm, there is already a link to "CVS repository" there, do you think
> we should make the link more prominent?

Whups, so there is... I thought that went somewhere else.  Already you
can see how reliable and error-free I am ;o)   Perhaps pointing this to
cvswrite/checkin.html might be more useful.


> >>  . Should we unify some of these pages? For example, cvs.html and
> >>    cvswrite.html?
> > I think it would be a good idea to merge cvswrite.html and
> > contribute.html, which cover similar territory.
> 
> Sounds like a good idea to me!
> 
> But you did mean to write cvswrite.html and checkin.html, didn't you?

Yes *sigh*  :o)

 
> > That's me :o)  I find it easy enough to follow.  One thing I did find
> > a wee bit excessive was the description of a cvs check-in under
> > "Example check-in session" in checkin.html.  While this was very
> > useful, I suggest providing a separate page and linking to it.  But
> > then, my preference for web pages is bite-sized pieces [...]
> 
> Perhaps we can keep it at the very end of the unified
> cvswrite+contribute.html page and add a small index at the beginning
> of the page, something like the following?
> 
>  <A HREF="heading1">Heading 1</A> | <A HREF="heading2">Heading 2</A>...

Sure, I'll take care of that.

 
> > A general comment on presentation: I think it would make sense for
> > these pages to move towards white backgrounds, black text.  Some are,
> > some aren't, and it looks wierd.
> 
> Yes, yes, and yes. Indeed that's what I am currently working on. :-)
>   http://gcc.gnu.org/ml/gcc-patches/1999-12/msg00968.html
> 
> Let's make a deal: I take care of this issue, and you attack our cvs
> web pages. ;-)

:o)  It's a deal.

Cheers,

--
Rich
churchmouse@mpx.com.au

Re: GCC museum?

[Rich Churcher]

Gerald Pfeifer writes:
 > On Sun, 19 Dec 1999, Rich Churcher wrote:
 > > > PS: And sorry for the delay, but -- finally -- my GCC web pages TODO list
 > > > is getting smaller and smaller. ;-)
 > > If you need any help whittling it down, please let me know :o)
 > 
 > Well, you asked for it. :-)

That's ok, I meant it :o) I enjoyed helping out the last time (with
the FAQ), but due to moving countries and jobs I never got around to
writing back and asking for more "assignments".  I use GCC every day,
and always feel like I should be contributing more to the project.


 > It would be great if you could have a close look at those pages linked to
 > under the "Development" heading of our main page, especially contribute.html,
 > cvs.html, cvswrite.html, and checkin.html.
 > 
 >  . Can we compress the information there a bit?
 > 
 >  . Should we add further information?
 > 
 >  . Do we need a couple of further cross-links between these pages?
 > 
 >  . Should we unify some of these pages? For example, cvs.html and
 >    cvswrite.html?
 > 
 >  . Is the information there clear for a non-expert?
 >    Is it easy enough for an expert to find relevant information quickly?

I'll have a look and let you know.

-- 
Rich
churchmouse@mpx.com.au

Web pages (was: GCC museum?)

[Gerald Pfeifer]

On Thu, 30 Dec 1999, Rich Churcher wrote:
> A link to cvs.html under the Submitting Patches heading in
> contribute.html wouldn't hurt.

Hmm, there is already a link to "CVS repository" there, do you think
we should make the link more prominent?

(Some people still use snapshots and submit patches based one these.)

>>  . Should we unify some of these pages? For example, cvs.html and
>>    cvswrite.html?
> I think it would be a good idea to merge cvswrite.html and
> contribute.html, which cover similar territory.

Sounds like a good idea to me!

But you did mean to write cvswrite.html and checkin.html, didn't you?

> cvs.html is sort of different, in that you could have read access
> purely as a matter of interest, without needing to know about write
> access and procedures.

Good point!

> That's me :o)  I find it easy enough to follow.  One thing I did find
> a wee bit excessive was the description of a cvs check-in under
> "Example check-in session" in checkin.html.  While this was very
> useful, I suggest providing a separate page and linking to it.  But
> then, my preference for web pages is bite-sized pieces [...]

Perhaps we can keep it at the very end of the unified
cvswrite+contribute.html page and add a small index at the beginning
of the page, something like the following?

 <A HREF="heading1">Heading 1</A> | <A HREF="heading2">Heading 2</A>...

>> Is it easy enough for an expert to find relevant information quickly?
> Again, how would I know?  Ok, I'm being obtuse.  Let me put it another 
> way - I suspect that experts will already know the logistics, and just 
> need a few details.  If you like, I can put those at the top and
> further explanations further down on each page.  Is that the kind of
> structure that would make sense to a community like GCC?  Or a link at 
> the top of the page marked "Essential details", or some such.

First the general stuff and then the those details would be my preference.

Essentially the "Setting up your local CVS tree" section of cvswrite.html
already contains the important information for experts.

> A general comment on presentation: I think it would make sense for
> these pages to move towards white backgrounds, black text.  Some are,
> some aren't, and it looks wierd.

Yes, yes, and yes. Indeed that's what I am currently working on. :-)
  http://gcc.gnu.org/ml/gcc-patches/1999-12/msg00968.html

Let's make a deal: I take care of this issue, and you attack our cvs
web pages. ;-)

> Finally - should I be discussing this on gcc-patches?  Sorry, it's
> been awhile...

Before there's a patch, this list is apropriate, AFAIS. But I'm looking
forward to receiving a patch via gcc-patches! :-) 

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/

Re: GCC museum?

[Rich Churcher]

Gerald Pfeifer writes:
 > 
 > It would be great if you could have a close look at those pages linked to
 > under the "Development" heading of our main page, especially contribute.html,
 > cvs.html, cvswrite.html, and checkin.html.
 > 
 >  . Can we compress the information there a bit?

I've had a chance to look over these pages.  Actually, everything
seems solid and well-organised, so any changes will be incremental.

 > 
 >  . Should we add further information?

This is perhaps not the best question for me to answer.  I qualify for 
the "non-expert" point of view, as referenced below :o)  However, I
promise faithfully to add anything others consider relevant into any
patch I submit.  It looks complete to my untrained eyes.

 
 >  . Do we need a couple of further cross-links between these pages?

Maybe.  A link to cvs.html under the Submitting Patches heading in
contribute.html wouldn't hurt.  If you're going to merge the cvs
pages, you don't need much else.


 >  . Should we unify some of these pages? For example, cvs.html and
 >    cvswrite.html?

I think it would be a good idea to merge cvswrite.html and
contribute.html, which cover similar territory.  cvs.html is sort of
different, in that you could have read access purely as a matter of
interest, without needing to know about write access and procedures.


 >  . Is the information there clear for a non-expert?

That's me :o)  I find it easy enough to follow.  One thing I did find
a wee bit excessive was the description of a cvs check-in under
"Example check-in session" in checkin.html.  While this was very useful, I 
suggest providing a separate page and linking to it.  But then, my
preference for web pages is bite-sized pieces rather than lots of
scrolling - YMMV.


 >    Is it easy enough for an expert to find relevant information
 > quickly?

Again, how would I know?  Ok, I'm being obtuse.  Let me put it another 
way - I suspect that experts will already know the logistics, and just 
need a few details.  If you like, I can put those at the top and
further explanations further down on each page.  Is that the kind of
structure that would make sense to a community like GCC?  Or a link at 
the top of the page marked "Essential details", or some such.

A general comment on presentation: I think it would make sense for
these pages to move towards white backgrounds, black text.  Some are,
some aren't, and it looks wierd.  I presume some may have been
automatically generated from .texi, or whatever.  Still - compare
projects.html with cvs.html, for example.

Finally - should I be discussing this on gcc-patches?  Sorry, it's
been awhile...

Cheers,

-- 
Rich
churchmouse@mpx.com.au

Re: GCC museum?

[Rich Churcher]

Gerald Pfeifer writes:
 > 
 > PS: And sorry for the delay, but -- finally -- my GCC web pages TODO list
 > is getting smaller and smaller. ;-)

If you need any help whittling it down, please let me know :o)

Cheers,

-- 
Rich
churchmouse@mpx.com.au