ACATS results with Gnat5.01p

ACATS results with Gnat5.01p

[Tapani Rundgren]

Gnat version GNATLS 5.01p (20031017)
Redhat Linux 9 with kernel 2.4.20-20.9
$ cat run.log |  grep 'acats4gnat result'
acats4gnat results cz 3 /       4
acats4gnat results a 75 /      75
acats4gnat results c2 34 /      34
acats4gnat results c3 353 /     356
acats4gnat results c4 340 /     340
acats4gnat results c5 95 /      95
acats4gnat results c6 81 /      81
acats4gnat results c7 53 /      53
acats4gnat results c8 141 /     141
acats4gnat results c9 255 /     255
acats4gnat results ca 74 /      74
acats4gnat results cb 43 /      43
acats4gnat results cc 117 /     118
acats4gnat results cd 177 /     178
acats4gnat results ce 265 /     265
acats4gnat results cxa 87 /      87
acats4gnat results cxb 30 /      30
acats4gnat results cxc 13 /      13
acats4gnat results cxd 32 /      39
acats4gnat results cxe 1 /       1
acats4gnat results cxf 20 /      20
acats4gnat results cxg 29 /      29
acats4gnat results cxh 4 /       4
acats4gnat results d 4 /       4
acats4gnat results e 11 /      11
acats4gnat results l 26 /      26

-- 
Rundgren dot Tapani at telia dot com

Re: ACATS results with Gnat5.01p

[Jeff C,]

"Tapani Rundgren" <NoSpam@telia.com> wrote in message
news:EmMnb.35427$dP1.116305@newsc.telia.net...
> Gnat version GNATLS 5.01p (20031017)
> Redhat Linux 9 with kernel 2.4.20-20.9
> $ cat run.log |  grep 'acats4gnat result'

...stuff deleted.


Thanks...That is useful. I had wanted to run ACATS before I released the RPM
but I knew I'd never
get to it. Do you have similar results for the vanilla gcc 3.3.x series. It
would be interesting to see if 5.01p actually
performs better.

I have found a few cases where the vanilla gcc 3.3.x series compiler would
crash on valid code ant the 5.01p
compiled it successfully but I have done very little in the way of proving
the correctness of the generated code.

By the way, in case anyone is wondering about what we are calling 5.01p here
you can get it from
(Preferably) Adaworld by going to the
http://www.adaworld.com/linksmain.html
page and scrolling down.

If you want to download it from a much much slower site you can go to my
server
http://newserver.thecreems.com/index.php?topic=Programming

There is a little write-up on this site about the way this was put together.

Re: ACATS results with Gnat5.01p

[Tapani Rundgren]

Jeff C, wrote:
> "Tapani Rundgren" <NoSpam@telia.com> wrote in message
> news:EmMnb.35427$dP1.116305@newsc.telia.net...
> 
>>Gnat version GNATLS 5.01p (20031017)
>>Redhat Linux 9 with kernel 2.4.20-20.9
>>$ cat run.log |  grep 'acats4gnat result'
> 
> 
> ...stuff deleted.
> 
> 
> Thanks...That is useful. I had wanted to run ACATS before I released the RPM
> but I knew I'd never
> get to it. Do you have similar results for the vanilla gcc 3.3.x series. It
> would be interesting to see if 5.01p actually
> performs better.
> 
Just built Gcc-3.4 -> 20031022 snaptshot
Will test with aws-1.3 first
> I have found a few cases where the vanilla gcc 3.3.x series compiler would
> crash on valid code ant the 5.01p
> compiled it successfully but I have done very little in the way of proving
> the correctness of the generated code.
> 
> By the way, in case anyone is wondering about what we are calling 5.01p here
> you can get it from
> (Preferably) Adaworld by going to the
> http://www.adaworld.com/linksmain.html
> page and scrolling down.
> 
> If you want to download it from a much much slower site you can go to my
> server
> http://newserver.thecreems.com/index.php?topic=Programming
> 
> There is a little write-up on this site about the way this was put together.
> 
> 
> 
Well, aws-1.3 did build (no xmlada yet), but did not work!
Will run acats test now!

-- 
Rundgren dot Tapani at telia dot com

Re: ACATS results with Gnat5.01p

[Duncan Sands]

Sorry, but what is gnat5.01p?  Is it based on one
of ACT's builds?

Thanks,

Duncan.

Re: ACATS results with Gnat5.01p

[Jeff C,]

"Duncan Sands" <baldrick@free.fr> wrote in message
news:mailman.240.1067440519.25614.comp.lang.ada@ada-france.org...
> Sorry, but what is gnat5.01p?  Is it based on one
> of ACT's builds?
>
> Thanks,
>
> Duncan.

Somewhat.. Until recently the semi-internal ACT CVS archive for the Ada
directory was
available online along with build instructions. What was apparently in the
CVS was the actual XX.XXa series
sources (what gets built and shipped to the customers). the 5.01p that I
built was originally based on the released CVS tag of 5.01a BUT this by no
means implies that what I ended up with was 5.01a!

There are several reasons for that.

1) I modified the version file so that it would not call itself 5.01a. It
not identifies itself as 5.01p with a build date appended.

2) I built with gcc 3.2.2. I THINK that they were building with 3.2 (no
additional .2)

3) It is unclear/unknown to me if they every supported building inder Redhat
9.

4) There build instructions for building from source indicates that you had
to reuse a portion of
  your existing binary install to get a few components that I deduced may
have come from binutils so I built
  them myself.

5) In order to build the binutils under redhat 9, I probably had to use a
newer version they they did/would have.

6) I had no configuration manager / QA making sure I was even really
following the instructions that I posted.

7) I did essentially zero true functional testing of the resulting build

8) I bundled it as an RPM. No prior semi-official ACT releases that I am
aware of were done that way (not that there
was the Ada for Linux team that built RPMs and the compiler from
source...Note that they also just called theirs 3.14p (or whatever version
they were trying to build).

9) Many more items that I probably am not even aware of.


You can read more at

http://newserver.thecreems.com/index.php?topic=Programming

If you feel the need to actually download stuff, try to get it from the
http://www.adaworld.com/linksmain.html
site since I do not really have any substantial upload capability.

Re: ACATS results with Gnat5.01p

[Duncan Sands]

Hi Jeff, thanks for doing this.

> 7) I did essentially zero true functional testing of the resulting build

It crashed on the first program I tried it on...  Oh well :)

Duncan.

Re: ACATS results with Gnat5.01p

[Jeff C,]

"Duncan Sands" <baldrick@free.fr> wrote in message
news:mailman.243.1067499423.25614.comp.lang.ada@ada-france.org...
> Hi Jeff, thanks for doing this.
>
> > 7) I did essentially zero true functional testing of the resulting build
>
> It crashed on the first program I tried it on...  Oh well :)
>
> Duncan.

Note that 0 testing still means it was capable enough to compile itself (via
the gcc bootstrap process) and
then compile the GNAT runtime libraries. Since it bootstraps the compiler it
builds with itself is then also
run so it certainly is not totally broken (also seems to have passed (at
this point) most of the ACATS suite).

Also successfully compiles GtkAda and the GtkAda test program. (Which then
runs fine)

So, while there probably are quite a few things wrong with the packaging of
it (for example, it should probably
really update /etc/ld.so.conf so you don't have to to get things like GtkAda
to work) it certainly
is not totally broken.

Re: ACATS results with Gnat5.01p

[Stephen Leake]

"Jeff C," <nolongersafeto@userealemailsniff.com> writes:

> I had wanted to run ACATS before I released the RPM but I knew I'd
> never get to it. Do you have similar results for the vanilla gcc
> 3.3.x series. It would be interesting to see if 5.01p actually
> performs better.

Can I request that you _not_ call your version of GNAT 5.01p? That's
the name ACT will use for their public release when they get around to
it. 

Please call yours 5.01j (for Jeff), or something. Adding the date is
not enough.

-- 
-- Stephe

Re: ACATS results with Gnat5.01p

[Stephane Richard]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 877 bytes --]

Well that version number was proposed by ACT itself Steph after Jeff first
named it 5.01a ;-).  they talked on the phone and ACT told him he could call
it 5.01p :-)

-- 
St�phane Richard
"Ada World" Webmaster
http://www.adaworld.com


"Stephen Leake" <Stephe.Leake@nasa.gov> wrote in message
news:uznfkszm3.fsf@nasa.gov...
> "Jeff C," <nolongersafeto@userealemailsniff.com> writes:
>
> > I had wanted to run ACATS before I released the RPM but I knew I'd
> > never get to it. Do you have similar results for the vanilla gcc
> > 3.3.x series. It would be interesting to see if 5.01p actually
> > performs better.
>
> Can I request that you _not_ call your version of GNAT 5.01p? That's
> the name ACT will use for their public release when they get around to
> it.
>
> Please call yours 5.01j (for Jeff), or something. Adding the date is
> not enough.
>
> -- 
> -- Stephe

Re: ACATS results with Gnat5.01p

[Stephen Leake]

"Stephane Richard" <stephane.richard@verizon.net> writes:

> Well that version number was proposed by ACT itself Steph after Jeff first
> named it 5.01a ;-).  they talked on the phone and ACT told him he could call
> it 5.01p :-)

Hmm. Ok, I guess that means ACT won't be releasing 5.01p; they'll just
be updating the FSF tree. I guess that's a Good Thing.

-- 
-- Stephe