Configure.ac patch for AIX 5.3

[fish-8]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Alexey wrote:

> to produce 64-bit Herc executable:
>
> ./configure CC='gcc -maix64' NM='nm -B -X64' AR='ar -X64' LDFLAGS='-
> L/usr/local/lib' CFLAGS='-I/usr/local/include -D_LARGE_FILES'
>
> to produce 32-bit Herc executable:
>
> ./configure CC='gcc' LDFLAGS='-L/usr/local/lib'
CFLAGS='-I/usr/local/include
> -D_LARGE_FILES'

and:

> to produce 64-bit Herc executable:
>
> ./configure CC='xlC -q64' CXX='xlC -q64' NM='nm -B -X64' AR='ar -X64'
> LDFLAGS='-L/usr/local/lib' CFLAGS='-I/usr/local/include -D__POWERPC__ -
> D_LARGE_FILES'
>
> to produce 32-bit Herc executable:
>
> ./configure CC='xlC' CXX='xlC' LDFLAGS='-L/usr/local/lib' CFLAGS='-
> I/usr/local/include -D__POWERPC__ -D_LARGE_FILES'

OKAY...

Here's my first attempt at fixing configure for AIX. Please apply the below
patch and try both gcc and IBM VA C/xlC again.

HOPEFULLY they both should now work "out of the box" without *any* other
changes (except the stuff in hostopts.h of course).

The only thing that still won't work (which is LAST on my/our list of
"things to do") is the SCSI tape issue. I'm not concerned about that right
now. I'm concentrating on getting all the other stuff resolved first.

With the below patch applied you should NOT have to specify ANY special
configure parameters. It should automatically detect a 64-bit build and
automatically add the needed -q64 or -maix64 options, etc. Just run
"./configure" without any parameters at all.

The ONLY thing you WILL need to do beforehand is "export OBJECT_MODE=64" if
you want to build a 64-bit Hercules or "export OBJECT_MODE=32" (or don't
export at all) if you want to build a 32-bit Hercules.
 
All of the below changes are based on various web pages I've come across
during my research so it still may not work right (especially given that I'm
a Windoze weenie and not a *nix person nor very skilled at autoconf and
bash, etc).

Please apply it against current SVN and tell me how it goes. Thanks.

(If it's still not right I'm hoping it's at least close!)

Here's the patch:


   ftp://www.softdevlabs.com/fish/pub/Alexey/aix.pat.v1.txt


Thanks.

- --
"Fish" (David B. Trout) - fish@...
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52




-----BEGIN PGP SIGNATURE-----
Version: PGP Desktop 9.8.0 (Build 2158)
Charset: us-ascii

wj8DBQFK52gBSPXX9MTuPioRAkyLAKDldSNdLu4mMAjSvLONvkO4NKaQbwCfYqQU
oy7A3sKvP1cVQu1L10R5JhY=
=hFpG
-----END PGP SIGNATURE-----

[Harold Grovesteen-2]

Fish wrote:

>
>Here's my first attempt at fixing configure for AIX. Please apply the below
>patch and try both gcc and IBM VA C/xlC again.
>
>HOPEFULLY they both should now work "out of the box" without *any* other
>changes (except the stuff in hostopts.h of course).
>
>The only thing that still won't work (which is LAST on my/our list of
>"things to do") is the SCSI tape issue. I'm not concerned about that right
>now. I'm concentrating on getting all the other stuff resolved first.
>
>With the below patch applied you should NOT have to specify ANY special
>configure parameters. It should automatically detect a 64-bit build and
>automatically add the needed -q64 or -maix64 options, etc. Just run
>"./configure" without any parameters at all.
>
>The ONLY thing you WILL need to do beforehand is "export OBJECT_MODE=64" if
>you want to build a 64-bit Hercules or "export OBJECT_MODE=32" (or don't
>export at all) if you want to build a 32-bit Hercules.
>

Is there anyway to determine OBJECT_MODE from the version of AIX. The
AIX releases are tied to the underlying version of the POWER chip.
Should that not give us a clue of whether 64 or 32 bit executable should
be built?

>
>All of the below changes are based on various web pages I've come across
>during my research so it still may not work right (especially given that I'm
>a Windoze weenie and not a *nix person nor very skilled at autoconf and
>bash, etc).
>
>Please apply it against current SVN and tell me how it goes. Thanks.
>
>(If it's still not right I'm hoping it's at least close!)
>
>Here's the patch:
>
>
>   ftp://www.softdevlabs.com/fish/pub/Alexey/aix.pat.v1.txt
>

I see we are moving from using host to target system for this detection.
In a number of places scattered throughout configure.ac there are other
references to the host. Should these change to target? If host=target,
why do we need to change to target? Are we trying to allow
cross-compilation of Hercules?

So, was the real configure problem that we simply did not have proper
detection for aix and defaulted to gcc?

>
>
>Thanks.
>
>- --
>"Fish" (David B. Trout) - fish@...
>Fight Spam! Join CAUCE! <http://www.cauce.org/>
>7 reasons why HTML email is a bad thing
>http://www.georgedillon.com/web/html_email_is_evil.shtml
>PGP key fingerprints:
>DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A
>RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52
>
>
>
>
>-----BEGIN PGP SIGNATURE-----
>Version: PGP Desktop 9.8.0 (Build 2158)
>Charset: us-ascii
>
>wj8DBQFK52gBSPXX9MTuPioRAkyLAKDldSNdLu4mMAjSvLONvkO4NKaQbwCfYqQU
>oy7A3sKvP1cVQu1L10R5JhY=
>=hFpG
>-----END PGP SIGNATURE-----
>
>
>------------------------------------
>
>Community email addresses:
>  Post message: hercules-390@...
>  Subscribe:    hercules-390-subscribe@...
>  Unsubscribe:  hercules-390-unsubscribe@...
>  List owner:   hercules-390-owner@...
>
>Files and archives at:
>  http://groups.yahoo.com/group/hercules-390
>
>Get the latest version of Hercules from:
>  http://www.hercules-390.org
>
>Yahoo! Groups Links
>
>
>
>
>  
>

[fish-8]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Harold Grovesteen wrote:

[...]
> I see we are moving from using host to target system for
> this detection. In a number of places scattered throughout
> configure.ac there are other references to the host. Should
> these change to target?

Yes and they already have been. I just didn't bother including those changes
in the patch.


> If host=target, why do we need to change to target? Are we
> trying to allow cross-compilation of Hercules?

Sure, why not?
 

> So, was the real configure problem that we simply did not
> have proper detection for aix and defaulted to gcc?

Kind of. We were presuming that gcc was always going to be used, and thus
were always initializing CFLAGS to "-W -Wall" (to prevent AC_PROG_CC from
initializing it to "-g -O2" which we don't want since we prefer deciding for
ourselves which optimization flags to use). That was screwing things up for
Alexey whenever he tried using VisualAge. (VisualAge doesn't like/understand
"-W -Wall").

So since we were already doing some hard-coded detection for some other
platform I figured I'd add one for AIX and presume that if GCC wasn't being
used (on AIX) then they must be using VisualAge and then make the necessary
adjustments.

ANYWAY... I'm not sure if that answers your question or not but since Alexey
was always requesting 64-bit builds by always having to remember to always
manually specify the right CC/AR/NM/CFLAGS values on his configure command
(which is ugly and prone to user error) I thought I'd try to make things a
bit simpler and automate the process somewhat. Now all the user has to do is
export OBJECT_MODE=64 and then enter "configure" without any parameters and
our configure script [hopefully] does the right thing. <shrug>

(I was going to make the choice of building 64-bit or 32-bit a configurable
option but decided that could probably wait for some other day and so
decided to key off of the OBJECT_MODE variable instead since it appeared
everyone else was (including IBM's VisualAge))

As far as the cross-compilation thing goes (the change to testing target_os
instead of host_os like we were), I just figured it was just "more correct"
to do so. <shrug> In most situations they'd be the same since in most
situations the user would be doing the build on the same system s/he
intended to run the resulting binary on so the end effect of checking
target_xxx instead of host_xxx would end up being the same.

In those situations where they /weren't/ however, then checking host_os
instead of target_os would cause us to do the wrong thing, so even though it
hasn't been a problem for us YET, I figured why wait to fix it? If we know
something is technically wrong, why not try to fix it NOW than run the risk
of it coming back to bite us on the *ss later?

Have you heard from Alexey yet? I'm dying to know whether my proposed patch
works for him or not!

- --
"Fish" (David B. Trout) - fish@...
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52





-----BEGIN PGP SIGNATURE-----
Version: PGP Desktop 9.8.0 (Build 2158)
Charset: us-ascii

wj8DBQFK6I0gSPXX9MTuPioRAi8LAJwOMN9EfZhMptpRTXnX4XU2SZnwiwCg5eRn
h7PWMyxmVkJwU9WzjeW644U=
=KzbJ
-----END PGP SIGNATURE-----

[Harold Grovesteen-2]

Fish wrote:

>-----BEGIN PGP SIGNED MESSAGE-----
>Hash: SHA1
>
>Harold Grovesteen wrote:
>
>[...]
>  
>
>>I see we are moving from using host to target system for
>>this detection. In a number of places scattered throughout
>>configure.ac there are other references to the host. Should
>>these change to target?
>>    
>>
>
>Yes and they already have been. I just didn't bother including those changes
>in the patch.
>  
>

Good. The inconsistency bothered me.

>
>  
>
>>If host=target, why do we need to change to target? Are we
>>trying to allow cross-compilation of Hercules?
>>    
>>
>
>Sure, why not?
>
>
>  
>
>>So, was the real configure problem that we simply did not
>>have proper detection for aix and defaulted to gcc?
>>    
>>
>
>Kind of. We were presuming that gcc was always going to be used, and thus
>were always initializing CFLAGS to "-W -Wall" (to prevent AC_PROG_CC from
>initializing it to "-g -O2" which we don't want since we prefer deciding for
>ourselves which optimization flags to use). That was screwing things up for
>Alexey whenever he tried using VisualAge. (VisualAge doesn't like/understand
>"-W -Wall").
>
>So since we were already doing some hard-coded detection for some other
>platform I figured I'd add one for AIX and presume that if GCC wasn't being
>used (on AIX) then they must be using VisualAge and then make the necessary
>adjustments.
>
>ANYWAY... I'm not sure if that answers your question or not but since Alexey
>was always requesting 64-bit builds by always having to remember to always
>manually specify the right CC/AR/NM/CFLAGS values on his configure command
>(which is ugly and prone to user error) I thought I'd try to make things a
>bit simpler and automate the process somewhat. Now all the user has to do is
>export OBJECT_MODE=64 and then enter "configure" without any parameters and
>our configure script [hopefully] does the right thing. <shrug>
>

This still requires the end user to do something for what I suspect is
the default. Unless one is running a very old version of AIX, 64-bit is
what you would want. Can we make 64-bit the default without precluding
32-bit as an option if needed. Alexy or others who are more familiar
with AIX should chime in on this. My thought was that based upon the
detected version of AIX, the decision could be made, but I don't know
enough on this topic to know how to make that decision in the code.

>
>(I was going to make the choice of building 64-bit or 32-bit a configurable
>option but decided that could probably wait for some other day and so
>decided to key off of the OBJECT_MODE variable instead since it appeared
>everyone else was (including IBM's VisualAge))
>
>As far as the cross-compilation thing goes (the change to testing target_os
>instead of host_os like we were), I just figured it was just "more correct"
>to do so. <shrug> In most situations they'd be the same since in most
>situations the user would be doing the build on the same system s/he
>intended to run the resulting binary on so the end effect of checking
>target_xxx instead of host_xxx would end up being the same.
>

I think really cross-compiling Hercules is a much more complex task and
the configure.ac code is not really prepared to do that. Use of target
instead of host is definitely the right direction for it though.

>
>In those situations where they /weren't/ however, then checking host_os
>instead of target_os would cause us to do the wrong thing, so even though it
>hasn't been a problem for us YET, I figured why wait to fix it? If we know
>something is technically wrong, why not try to fix it NOW than run the risk
>of it coming back to bite us on the *ss later?
>
Yup. See previous comment.

>
>Have you heard from Alexey yet? I'm dying to know whether my proposed patch
>works for him or not!
>
No. I am anxious to here as well.

BTW, on the developers list there were some discussions about this and
any SVN commits for these changes need to go in after 3.07 is released.
Testing these changes with other NIX platforms needs to happen before we
include them in the release. I am sure everyone is preferring we not
have a 3.07.1 release a month after the 3.07 release is made available.

>
>- --
>"Fish" (David B. Trout) - fish@...
>Fight Spam! Join CAUCE! <http://www.cauce.org/>
>7 reasons why HTML email is a bad thing
>http://www.georgedillon.com/web/html_email_is_evil.shtml
>PGP key fingerprints:
>DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A
>RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52
>
>
>
>
>
>-----BEGIN PGP SIGNATURE-----
>Version: PGP Desktop 9.8.0 (Build 2158)
>Charset: us-ascii
>
>wj8DBQFK6I0gSPXX9MTuPioRAi8LAJwOMN9EfZhMptpRTXnX4XU2SZnwiwCg5eRn
>h7PWMyxmVkJwU9WzjeW644U=
>=KzbJ
>-----END PGP SIGNATURE-----
>
>
>------------------------------------
>
>Community email addresses:
>  Post message: hercules-390@...
>  Subscribe:    hercules-390-subscribe@...
>  Unsubscribe:  hercules-390-unsubscribe@...
>  List owner:   hercules-390-owner@...
>
>Files and archives at:
>  http://groups.yahoo.com/group/hercules-390
>
>Get the latest version of Hercules from:
>  http://www.hercules-390.org
>
>Yahoo! Groups Links
>
>
>
>
>  
>

[fish-8]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Harold Grovesteen wrote:

<snip; export OBJECT_MODE=64>

> This still requires the end user to do something for what
> I suspect is the default. Unless one is running a very old
> version of AIX, 64-bit is what you would want. Can we make
> 64-bit the default without precluding 32-bit as an option
> if needed.

You're probably right that we should make 64-bit the default and only build
32-bit if, say, OBJECT_MODE=32 is set.


> Alexey or others who are more familiar with AIX should chime
> in on this.

Agreed.


> My thought was that based upon the detected version of AIX,
> the decision could be made, but I don't know enough on this
> topic to know how to make that decision in the code.

Hmmm... Yes... Completely overlooked that. My bad.

I should probably be checking whether target_os == host_os and if so, trying
to determine on my own whether the host system is 64-bit or not.

But of course in the cross compiling situation (target_os != host_os) then
setting OBJECT_MODE would need to be required (until such time as I or
someone else can get energetic enough to add a "--64-bit" option (or
something similar) to our configure.ac script that is).
 

[...]
> BTW, on the developers list there were some discussions
> about this and any SVN commits for these changes need to
> go in after 3.07 is released.

10-4.  I'll hold off on committing anything.


> Testing these changes with other NIX platforms needs to
> happen before we include them in the release. I am sure
> everyone is preferring we not have a 3.07.1 release a
> month after the 3.07 release is made available.

Agreed.

3.07.1 should be released *TWO* months after 3.07.

<rim shot>  [*]

   ;-)

- --
"Fish" (David B. Trout) - fish@...
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52


[*] http://www.youtube.com/watch?v=E0PIdWdw15U




-----BEGIN PGP SIGNATURE-----
Version: PGP Desktop 9.8.0 (Build 2158)
Charset: us-ascii

wj8DBQFK6a3ESPXX9MTuPioRAqgtAJwOEt49Za+y0YZnood/EEqEFSDQugCg/ruk
aPr6mcWSulvr7KXsL9FMgls=
=4c7i
-----END PGP SIGNATURE-----

[Alexey Bozrikov]

Dear Alexey,

Friday, 30 October 2009, you wrote:

HG>> Fish wrote:

>>>
>>>Here's my first attempt at fixing configure for AIX. Please apply the below
>>>patch and try both gcc and IBM VA C/xlC again.
>>>
>>>HOPEFULLY they both should now work "out of the box" without *any* other
>>>changes (except the stuff in hostopts.h of course).
>>>
>>>The only thing that still won't work (which is LAST on my/our list of
>>>"things to do") is the SCSI tape issue. I'm not concerned about that right
>>>now. I'm concentrating on getting all the other stuff resolved first.
>>>
>>>With the below patch applied you should NOT have to specify ANY special
>>>configure parameters. It should automatically detect a 64-bit build and
>>>automatically add the needed -q64 or -maix64 options, etc. Just run
>>>"./configure" without any parameters at all.
>>>
>>>The ONLY thing you WILL need to do beforehand is "export OBJECT_MODE=64" if
>>>you want to build a 64-bit Hercules or "export OBJECT_MODE=32" (or don't
>>>export at all) if you want to build a 32-bit Hercules.
>>>

HG>> Is there anyway to determine OBJECT_MODE from the version of AIX. The
HG>> AIX releases are tied to the underlying version of the POWER chip.
HG>> Should that not give us a clue of whether 64 or 32 bit executable should
HG>> be built?

>>>
>>>All of the below changes are based on various web pages I've come across
>>>during my research so it still may not work right (especially given that I'm
>>>a Windoze weenie and not a *nix person nor very skilled at autoconf and
>>>bash, etc).
>>>
>>>Please apply it against current SVN and tell me how it goes. Thanks.
>>>
>>>(If it's still not right I'm hoping it's at least close!)
>>>
>>>Here's the patch:
>>>
>>>
>>>   ftp://www.softdevlabs.com/fish/pub/Alexey/aix.pat.v1.txt
>>>

A> Hmmm, somehow I did not receive the original message from Fish about
A> the patch. Anyway - the link is here, I downloaded the file. Seems to
A> be my dumbness - I cannot apply the patch. patch does not seem to
A> 'understand' this file:

A> $ patch -v
A> Header: patch.c,v 2.0.1.6 88/06/22 20:46:39 lwall Locked
A> Patch level: 12
A>  $ wget ftp://www.softdevlabs.com/fish/pub/Alexey/aix.pat.v1.txt
A> --09:52:15--  ftp://www.softdevlabs.com/fish/pub/Alexey/aix.pat.v1.txt
A>            => `aix.pat.v1.txt'
A> Length: unspecified [text/plain]
A>     [ <=>                                                          ] 7,317         --.--K/s
A> 09:52:18 (338.09 KB/s) - `aix.pat.v1.txt' saved [7317]

A> $ cd hercules
A> $ patch -i ../aix.pat.v1.txt configure.ac
A> Hmm...  I can't seem to find a patch in there anywhere.


Indeed was my dumbness. patch on AIX is way too old! Patched the file on
Linux - all worked fine.
Now configure script does not complain anymore when using IBM VisualAge
compiler.

One step closer!

Alexey




** Taxation WITH representation isn't so hot, either.



------------------------------------

Community email addresses:
  Post message: hercules-390@...
  Subscribe:    hercules-390-subscribe@...
  Unsubscribe:  hercules-390-unsubscribe@...
  List owner:   hercules-390-owner@...

Files and archives at:
  http://groups.yahoo.com/group/hercules-390

Get the latest version of Hercules from:
  http://www.hercules-390.org

Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/hercules-390/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/hercules-390/join
    (Yahoo! ID required)

<*> To change settings via email:
    mailto:hercules-390-digest@...
    mailto:hercules-390-fullfeatured@...

<*> To unsubscribe from this group, send an email to:
    hercules-390-unsubscribe@...

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/

[Jeff Sumner-2]

The AIX 'patch' utility is among the first to bypass with Gnu alternatives,
along with tar (AIX's won't handle unpacking gcc in some archives) and make
(often the FIRST that I build!).

What's fun about this discussion for me (I've not been building on AIX over
the past week- too busy, and the RS/6000 I have is just too darned slow) is
that I brought up Hercules on the Mac (Turnkey) and got to play around again
for familiarity.
What a fantastic thing done- this software is just amazing.

I love bringing a crusty mainframer over to my desk at work- show them the
3270 session, and just start asking them questions about how to do things.
One of the best way to unlock some of the most sarcastic, funny, and
experienced people around. Wonderful stories to be heard from a lifetime
spent between the raised floor and the console.

Anyway. Thank you for your work, folks.

J

>
>
> Indeed was my dumbness. patch on AIX is way too old! Patched the file on
> Linux - all worked fine.
> Now configure script does not complain anymore when using IBM VisualAge
> compiler.
>
> One step closer!
>
> Alexey
>
>
>

[Alexey Bozrikov]

JS> The AIX 'patch' utility is among the first to bypass with Gnu alternatives,
JS> along with tar (AIX's won't handle unpacking gcc in some archives) and make
JS> (often the FIRST that I build!).


Right! I had this (wrong) impression that 'patch' on AIX is coming
from GNU utilities (so-called 'AIX toolbox for Linux Applications' CD)
and is fairly recent. Apparently, it is not.



JS> What's fun about this discussion for me (I've not been building on AIX over
JS> the past week- too busy, and the RS/6000 I have is just too darned slow) is
JS> that I brought up Hercules on the Mac (Turnkey) and got to play around again
JS> for familiarity.
JS> What a fantastic thing done- this software is just amazing.

JS> I love bringing a crusty mainframer over to my desk at work- show them the
JS> 3270 session, and just start asking them questions about how to do things.
JS> One of the best way to unlock some of the most sarcastic, funny, and
JS> experienced people around. Wonderful stories to be heard from a lifetime
JS> spent between the raised floor and the console.

I scared a couple of co-eds of mine with these screens! They thought
the technology died a long ago. Apparently, people think that all
these 'VMware' and 'PowerVM' technologies are really new ideas..



JS> Anyway. Thank you for your work, folks.

JS> J

>>
>>
>> Indeed was my dumbness. patch on AIX is way too old! Patched the file on
>> Linux - all worked fine.
>> Now configure script does not complain anymore when using IBM VisualAge
>> compiler.
>>
>> One step closer!
>>
>> Alexey
>>
>>
>>

** Your type doesn't stay around long enough to stay your type.



------------------------------------

Community email addresses:
  Post message: hercules-390@...
  Subscribe:    hercules-390-subscribe@...
  Unsubscribe:  hercules-390-unsubscribe@...
  List owner:   hercules-390-owner@...

Files and archives at:
  http://groups.yahoo.com/group/hercules-390

Get the latest version of Hercules from:
  http://www.hercules-390.org

Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/hercules-390/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/hercules-390/join
    (Yahoo! ID required)

<*> To change settings via email:
    mailto:hercules-390-digest@...
    mailto:hercules-390-fullfeatured@...

<*> To unsubscribe from this group, send an email to:
    hercules-390-unsubscribe@...

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/

[fish-8]

Harold Grovesteen wrote:

[...]
> Is there anyway to determine OBJECT_MODE from the
> version of AIX. The AIX releases are tied to the
> underlying version of the POWER chip. Should that
> not give us a clue of whether 64 or 32 bit executable
> should be built?

Well I *was* actually going to try doing that but had trouble finding code
that did that and so fell back to just checking the OBJECT_MODE variable.

But then I tried a little harder to find the right(?) code and I think I may
have found it.

ALEXEY:

Please try v2 of my patch and tell me whether or not it works okay for you.
Thanks:


  (dir) ftp://www.softdevlabs.com/fish/pub/Alexey
  (file) ftp://www.softdevlabs.com/fish/pub/Alexey/aix.pat.v2.txt


NOTE: I've uploaded a README as well, which I shall reproduces here since
it's so short:


  AIX patches should always be applied to current SVN version.
  Remove any previous patch(es) first then apply the new patch.

  You may also need to run some sort of 'sed' command against
  these patch files too before running them through the patch
  utility since they're coming from Windows and end with CRLF
  instead of just LF.


Give it a try and let us know how it goes.

Thanks!

--
"Fish" (David B. Trout) - fish@...
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52

[fish-8]

Alexey wrote:

[...]
> [...]patch does not seem to 'understand' this file:
[...]
> $ cd hercules
> $ patch -i ../aix.pat.v1.txt configure.ac
> Hmm...  I can't seem to find a patch in there anywhere.

Maybe it's being caused by the file's line-endings being CRLF instead of the
expected(?) LF?? (just a guess)

If so then I can fix that and re-upload it easily enough. Is that what's
causing it?

--
"Fish" (David B. Trout) - fish@...
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52

[Jeff Sumner-2]

(hint:   sed 's/^m//' aix.pat.vi.txt)

The ^M is generated in Bash 4.0 (and ksh, the AIX default, for that  
matter) by "ctrl-v ctrl-m"

J
On Oct 30, 2009, at 8:19 PM, Fish wrote:

> Alexey wrote:
>
> [...]
> > [...]patch does not seem to 'understand' this file:
> [...]
> > $ cd hercules
> > $ patch -i ../aix.pat.v1.txt configure.ac
> > Hmm... I can't seem to find a patch in there anywhere.
>
> Maybe it's being caused by the file's line-endings being CRLF  
> instead of the
> expected(?) LF?? (just a guess)
>
> If so then I can fix that and re-upload it easily enough. Is that  
> what's
> causing it?
>
> --
>

[Non-text portions of this message have been removed]

[fish-8]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Alexey wrote:

<snip; 'patch' problem..>

> Indeed was my dumbness. patch on AIX is way too old!
> Patched the file on Linux - all worked fine. Now configure
> script does not complain anymore when using IBM VisualAge
> compiler.
>
> One step closer!

Excellent!

Did you see my post about v2 of my patch? Have you tried it yet?

I made another tweak that first checks whether a cross-compile is being done
or not and only checks the OBJECT_MODE variable if you are. Otherwise if
you're not (normal case) I ignore the OBJECT_MODE variable and attempt to
determine the system's current 32/64 "bitness" mode myself by checking the
"/unix" link instead.
 
That way you shouldn't have to do anything special at all. Just run
./configure and that's it.

Since I'm making all these [proposed] changes without having an AIX system
of my own to test them with I *really* appreciate you helping me test them
Alexey.

Thanks.

- --
"Fish" (David B. Trout) - fish@...
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52





-----BEGIN PGP SIGNATURE-----
Version: PGP Desktop 9.8.0 (Build 2158)
Charset: us-ascii

wj8DBQFK64gbSPXX9MTuPioRAtGSAKCUfqSYXTiqeNrgl+w9I+45jUAKQACgkj87
bFHL7wkpx44FWSbLwU2fQP8=
=0rSA
-----END PGP SIGNATURE-----

[Alexey Bozrikov]

F> -----BEGIN PGP SIGNED MESSAGE-----
F> Hash: SHA1

F> Alexey wrote:

F> <snip; 'patch' problem..>

>> Indeed was my dumbness. patch on AIX is way too old!
>> Patched the file on Linux - all worked fine. Now configure
>> script does not complain anymore when using IBM VisualAge
>> compiler.
>>
>> One step closer!

F> Excellent!

F> Did you see my post about v2 of my patch? Have you tried it yet?

I just tried to use patch V2 against freshly downloaded SVN version -
no luck so far, patches are being rejected. I'll read the thread
through - possibly something changed during weekend.


[quote]
$ patch -p0 -i ../aix.pat.v2.txt
patching file configure.ac
Hunk #1 FAILED at 9.
Hunk #2 FAILED at 244.
Hunk #3 FAILED at 351.
Hunk #4 FAILED at 427.
Hunk #5 FAILED at 546.
Hunk #6 FAILED at 830.
Hunk #7 FAILED at 890.
Hunk #8 FAILED at 940.
Hunk #9 FAILED at 1144.
Hunk #10 FAILED at 1173.
Hunk #11 FAILED at 1204.
Hunk #12 FAILED at 1495.
Hunk #13 FAILED at 1608.
Hunk #14 FAILED at 1734.
Hunk #15 FAILED at 1797.
Hunk #16 FAILED at 1809.
Hunk #17 FAILED at 1832.
Hunk #18 FAILED at 1990.
Hunk #19 FAILED at 2021.
19 out of 19 hunks FAILED -- saving rejects to file configure.ac.rej
patching file README.AIX
patching file hostopts.h
Hunk #1 FAILED at 125.
Hunk #2 FAILED at 145.
Hunk #3 FAILED at 175.
Hunk #4 FAILED at 193.
Hunk #5 FAILED at 211.
Hunk #6 FAILED at 247.
Hunk #7 FAILED at 270.
Hunk #8 FAILED at 313.
Hunk #9 FAILED at 349.
9 out of 9 hunks FAILED -- saving rejects to file hostopts.h.rej
$ patch -v
patch 2.5.9
Copyright (C) 1988 Larry Wall
Copyright (C) 2003 Free Software Foundation, Inc.

This program comes with NO WARRANTY, to the extent permitted by law.
You may redistribute copies of this program
under the terms of the GNU General Public License.
For more information about these matters, see the file named COPYING.

written by Larry Wall and Paul Eggert

[unquote]

F> I made another tweak that first checks whether a cross-compile is being done
F> or not and only checks the OBJECT_MODE variable if you are. Otherwise if
F> you're not (normal case) I ignore the OBJECT_MODE variable and attempt to
F> determine the system's current 32/64 "bitness" mode myself by checking the
F> "/unix" link instead.

"Bitness" check could be done on AIX, I think. This is what it shows on my system:

 $ ls -l /unix
lrwxrwxrwx    1 root     system           21 11 Apr 2005  /unix -> /usr/lib/boot/unix_64

F>  
F> That way you shouldn't have to do anything special at all. Just run
F> ./configure and that's it.

F> Since I'm making all these [proposed] changes without having an AIX system
F> of my own to test them with I *really* appreciate you helping me test them
F> Alexey.

F> Thanks.





** "Cry havoc and let slip the dogs of e-mail!"



------------------------------------

Community email addresses:
  Post message: hercules-390@...
  Subscribe:    hercules-390-subscribe@...
  Unsubscribe:  hercules-390-unsubscribe@...
  List owner:   hercules-390-owner@...

Files and archives at:
  http://groups.yahoo.com/group/hercules-390

Get the latest version of Hercules from:
  http://www.hercules-390.org

Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/hercules-390/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/hercules-390/join
    (Yahoo! ID required)

<*> To change settings via email:
    mailto:hercules-390-digest@...
    mailto:hercules-390-fullfeatured@...

<*> To unsubscribe from this group, send an email to:
    hercules-390-unsubscribe@...

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/

[Alexey Bozrikov]

Dear Alexey,

Monday, 02 November 2009, you wrote:

F>> -----BEGIN PGP SIGNED MESSAGE-----
F>> Hash: SHA1

F>> Alexey wrote:

F>> <snip; 'patch' problem..>

>>> Indeed was my dumbness. patch on AIX is way too old!
>>> Patched the file on Linux - all worked fine. Now configure
>>> script does not complain anymore when using IBM VisualAge
>>> compiler.
>>>
>>> One step closer!

F>> Excellent!

F>> Did you see my post about v2 of my patch? Have you tried it yet?

A> I just tried to use patch V2 against freshly downloaded SVN version -
A> no luck so far, patches are being rejected. I'll read the thread
A> through - possibly something changed during weekend.


A> [quote]
A> $ patch -p0 -i ../aix.pat.v2.txt
A> patching file configure.ac
A> Hunk #1 FAILED at 9.
A> Hunk #2 FAILED at 244.
A> Hunk #3 FAILED at 351.
A> Hunk #4 FAILED at 427.
A> Hunk #5 FAILED at 546.
A> Hunk #6 FAILED at 830.
A> Hunk #7 FAILED at 890.
A> Hunk #8 FAILED at 940.
A> Hunk #9 FAILED at 1144.
A> Hunk #10 FAILED at 1173.
A> Hunk #11 FAILED at 1204.
A> Hunk #12 FAILED at 1495.
A> Hunk #13 FAILED at 1608.
A> Hunk #14 FAILED at 1734.
A> Hunk #15 FAILED at 1797.
A> Hunk #16 FAILED at 1809.
A> Hunk #17 FAILED at 1832.
A> Hunk #18 FAILED at 1990.
A> Hunk #19 FAILED at 2021.
A> 19 out of 19 hunks FAILED -- saving rejects to file configure.ac.rej
A> patching file README.AIX
A> patching file hostopts.h
A> Hunk #1 FAILED at 125.
A> Hunk #2 FAILED at 145.
A> Hunk #3 FAILED at 175.
A> Hunk #4 FAILED at 193.
A> Hunk #5 FAILED at 211.
A> Hunk #6 FAILED at 247.
A> Hunk #7 FAILED at 270.
A> Hunk #8 FAILED at 313.
A> Hunk #9 FAILED at 349.
A> 9 out of 9 hunks FAILED -- saving rejects to file hostopts.h.rej
A> $ patch -v
A> patch 2.5.9
A> Copyright (C) 1988 Larry Wall
A> Copyright (C) 2003 Free Software Foundation, Inc.

A> This program comes with NO WARRANTY, to the extent permitted by law.
A> You may redistribute copies of this program
A> under the terms of the GNU General Public License.
A> For more information about these matters, see the file named COPYING.

A> written by Larry Wall and Paul Eggert

A> [unquote]

Ohhh.. I goofed again. I assumed that patch 2.5.9 on AIX would work
better than 2.5.4 on Linux. Nah! As it was correctly pointed out earlier,
the patch file is in 'DOS' format. Somehow patch on Linux gets
away with this issue, but not on AIX. After applying 'set
filemode=unix' to the patch in vim on AIX everything patched just
fine!

Now configure appends '-q64' and '-X64' flags where
appropriate (checking with IBM VA C compiler now). -D_LFS_LARGE_FILE
and -D_LARGE_FILES are here as well, good progress! I just specified
the compiler (CC=xlC) and c++ compiler (CXX=xlC).

There are few warnings during configure process. One may be related to
a typo in configure.ac (when defining NM):


checking command to parse  -X64 -B output from xlC -q64 object... failed
configure:6930: result: no
configure:6964: checking the maximum length of command line arguments
configure:7029: result: 8192
configure:7040: checking command to parse  -X64 -B output from xlC -q64 object
configure:7129: xlC -q64 -c   -I/usr/local/include -D_LARGE_FILES -D_LFS_LARGEFILE -D_LINUX_SOURCE_COMPAT conftest.c >&5
configure:7132: $? = 0
configure:7136:  -X64 -B conftest.o \| sed -n -e 's/^.*[        ]\([BCDT][BCDT]*\)[     ][      ]*\(\)\([_A-Za-z][_A-Za-z0-9]*\)$/\1 \2\3 \3/p' \> conftest.nm
./configure[7137]: -X64:  not found



Another warning spotted, but it does not seem to affect anything (I
did not try make install yet!):

config.status: WARNING:  'po/Makefile.in.in' seems to ignore the --datarootdir settin



F>> I made another tweak that first checks whether a cross-compile is being done
F>> or not and only checks the OBJECT_MODE variable if you are. Otherwise if
F>> you're not (normal case) I ignore the OBJECT_MODE variable and attempt to
F>> determine the system's current 32/64 "bitness" mode myself by checking the
F>> "/unix" link instead.

A> "Bitness" check could be done on AIX, I think. This is what it shows on my system:

A>  $ ls -l /unix
A> lrwxrwxrwx    1 root     system           21 11 Apr 2005  /unix -> /usr/lib/boot/unix_64

F>>  
F>> That way you shouldn't have to do anything special at all. Just run
F>> ./configure and that's it.

F>> Since I'm making all these [proposed] changes without having an AIX system
F>> of my own to test them with I *really* appreciate you helping me test them
F>> Alexey.

F>> Thanks.




** German is essentially a form of assembly language consisting entirely of far calls heavily accented with throaty guttural sounds. ---ilvi



------------------------------------

Community email addresses:
  Post message: hercules-390@...
  Subscribe:    hercules-390-subscribe@...
  Unsubscribe:  hercules-390-unsubscribe@...
  List owner:   hercules-390-owner@...

Files and archives at:
  http://groups.yahoo.com/group/hercules-390

Get the latest version of Hercules from:
  http://www.hercules-390.org

Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/hercules-390/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/hercules-390/join
    (Yahoo! ID required)

<*> To change settings via email:
    mailto:hercules-390-digest@...
    mailto:hercules-390-fullfeatured@...

<*> To unsubscribe from this group, send an email to:
    hercules-390-unsubscribe@...

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/

[Alexey Bozrikov]

F> -----BEGIN PGP SIGNED MESSAGE-----
F> Hash: SHA1

F> Alexey wrote:

F> <snip; 'patch' problem..>

>> Indeed was my dumbness. patch on AIX is way too old!
>> Patched the file on Linux - all worked fine. Now configure
>> script does not complain anymore when using IBM VisualAge
>> compiler.
>>
>> One step closer!

F> Excellent!

F> Did you see my post about v2 of my patch? Have you tried it yet?

F> I made another tweak that first checks whether a cross-compile is being done
F> or not and only checks the OBJECT_MODE variable if you are. Otherwise if
F> you're not (normal case) I ignore the OBJECT_MODE variable and attempt to
F> determine the system's current 32/64 "bitness" mode myself by checking the
F> "/unix" link instead.

One note on AIX and OBJECT_MODE. According to IBM documentation (and
some threads about building Perl for AIX), 64-bit executables can be
created and run on 64-bit platforms even with 32-bit kernel, thus
allowing Herc to have larger MAINSIZE. I do not know how to check
reliably for the 64-bit platform. AIX 'prtconf' command comes to mind,
at least it produces consistent results on AIX 5.1, 5.3 and 6.1:

$ prtconf -c
CPU Type: 64-bit


Alexey


F>  
F> That way you shouldn't have to do anything special at all. Just run
F> ./configure and that's it.

F> Since I'm making all these [proposed] changes without having an AIX system
F> of my own to test them with I *really* appreciate you helping me test them
F> Alexey.

F> Thanks.





** You've been leading a dog's life. Stay off the furniture.



------------------------------------

Community email addresses:
  Post message: hercules-390@...
  Subscribe:    hercules-390-subscribe@...
  Unsubscribe:  hercules-390-unsubscribe@...
  List owner:   hercules-390-owner@...

Files and archives at:
  http://groups.yahoo.com/group/hercules-390

Get the latest version of Hercules from:
  http://www.hercules-390.org

Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/hercules-390/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/hercules-390/join
    (Yahoo! ID required)

<*> To change settings via email:
    mailto:hercules-390-digest@...
    mailto:hercules-390-fullfeatured@...

<*> To unsubscribe from this group, send an email to:
    hercules-390-unsubscribe@...

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/

[fish-8]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Alexey wrote:

<snip; OBJECT_MODE vs. ls -l /unix>

> One note on AIX and OBJECT_MODE. According to IBM documentation
> (and some threads about building Perl for AIX), 64-bit executables
> can be created and run on 64-bit platforms even with 32-bit kernel,
> thus allowing Herc to have larger MAINSIZE. I do not know how to
> check reliably for the 64-bit platform. AIX 'prtconf' command comes
> to mind, at least it produces consistent results on AIX 5.1, 5.3
> and 6.1:
>
> $ prtconf -c
> CPU Type: 64-bit

Thanks! prtconf it is!


  (dir) ftp://www.softdevlabs.com/fish/pub/Alexey
  (file) ftp://www.softdevlabs.com/fish/pub/Alexey/aix.pat.v3.txt


p.s. the patch was uploaded using Unix-style line endings this time, so you
shouldn't have to do any kind of special fixing of it before attempting to
apply it. Hope that makes things easier on you and I apologize for the wrong
line ending format on the previous patches.


p.p.s how many "RE:" subject prefixes do we need?! Sheesh! ;-)

- --
"Fish" (David B. Trout) - fish@...
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52






-----BEGIN PGP SIGNATURE-----
Version: PGP Desktop 9.8.0 (Build 2158)
Charset: us-ascii

wj8DBQFK7/O9SPXX9MTuPioRAiZXAJ98Xx/Cf1bAjL3rGVrFt+OqdjivSACgkzGh
XXO0UN+1srK9l2TWrk1uBBA=
=kXvM
-----END PGP SIGNATURE-----

[fish-8]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Alexey wrote:

[...]
> There are few warnings during configure process. One may be
> related to a typo in configure.ac (when defining NM):

<snip details>

> Another warning spotted, but it does not seem to affect anything
> (I did not try make install yet!):

<snip remainder>

Not sure what the po/Makefile.in.in --datarootdir warning was all about, but
in v3 of my patch (available for ftp download now) I've hopefully fixed the
undefined $NM variable issue.

As mentioned in my other post I'm now also using prtconf to check for
"bitness" as well, so unless something else crops up I think we're pretty
much done here! I'll commit my changes as soon as 3.07 is released.

Thanks for everything Alexey.


p.s. Even though AIX SCSI Tape Support should be my next priority
(Hercules-wise) I'm currently questioning the need for the "sync"
instruction in the machdep.h assists, but we can talk about that later once
I've had time to throw together a simple test program. Thanks.

- --
"Fish" (David B. Trout) - fish@...
Fight Spam! Join CAUCE! <http://www.cauce.org/>
7 reasons why HTML email is a bad thing
http://www.georgedillon.com/web/html_email_is_evil.shtml
PGP key fingerprints:
DH/DSS: 9F9B BAB0 BA7F C458 1A89 FE26 48F5 D7F4 C4EE 3E2A
RSA: 6B37 7110 7201 9917 9B0D 99E3 55DB 5D58 FADE 4A52





-----BEGIN PGP SIGNATURE-----
Version: PGP Desktop 9.8.0 (Build 2158)
Charset: us-ascii

wj8DBQFK7/btSPXX9MTuPioRAlaHAJ96GVPmTwkbOrH48gb5ogEPLSPOsgCdFm5P
hxPGaUZ2qZ/dY0KeiBJAK7o=
=JKtB
-----END PGP SIGNATURE-----

[au1john]

p.p.s how many "RE:" subject prefixes do we need?! Sheesh! ;-)


That seems to be mail client dependent. Outlook 2003 seemed to not add any.

[Alexey Bozrikov]

F> Alexey wrote:

F> <snip; OBJECT_MODE vs. ls -l /unix>

>> One note on AIX and OBJECT_MODE. According to IBM documentation
>> (and some threads about building Perl for AIX), 64-bit executables
>> can be created and run on 64-bit platforms even with 32-bit kernel,
>> thus allowing Herc to have larger MAINSIZE. I do not know how to
>> check reliably for the 64-bit platform. AIX 'prtconf' command comes
>> to mind, at least it produces consistent results on AIX 5.1, 5.3
>> and 6.1:
>>
>> $ prtconf -c
>> CPU Type: 64-bit

F> Thanks! prtconf it is!


F>   (dir) ftp://www.softdevlabs.com/fish/pub/Alexey
F>   (file) ftp://www.softdevlabs.com/fish/pub/Alexey/aix.pat.v3.txt


F> p.s. the patch was uploaded using Unix-style line endings this time, so you
F> shouldn't have to do any kind of special fixing of it before attempting to
F> apply it. Hope that makes things easier on you and I apologize for the wrong
F> line ending format on the previous patches.

Somehow DOS format crept in again (could it be FTP server + myself
using wget, then some sort of translation in ASCII mode?). Never mind,
wonderful `tr' utility did the job.


Now everything configures and compiles without any additional tweaking
of  options (apart from specifying the compiler, of course - I checked
for both gcc and xlC).


There is one weird thing happening _only_ when you configure with xlC,
but it does not affect the build process later on:
[quote config.log]
configure:8148: checking if xlC -q64 supports -c -o file.o
configure:8176: xlC -q64 -c   -I/usr/local/include -D_LARGE_FILES -D_LFS_LARGEFILE -D_LINUX_SOURCE_COMPAT -o out/conftest2.o conftest.c >&5
xlC: 1501-220 current directory is not writable
configure:8180: $? = 0
[unquote]


Overall this  is  a really big step forward! I'll try to see what is happening
on AIX 5.1, too, as previous posters reported compile problems in this
environment.



F> p.p.s how many "RE:" subject prefixes do we need?! Sheesh! ;-)

I am using THEBAT! e-mail client, it seems adding an extra Re: every
time by default. I guess I found an option to stop this 'prefixing'.




** Millions of years ago, man climbed out of the slime. You want to join the party?



------------------------------------

Community email addresses:
  Post message: hercules-390@...
  Subscribe:    hercules-390-subscribe@...
  Unsubscribe:  hercules-390-unsubscribe@...
  List owner:   hercules-390-owner@...

Files and archives at:
  http://groups.yahoo.com/group/hercules-390

Get the latest version of Hercules from:
  http://www.hercules-390.org

Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/hercules-390/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/hercules-390/join
    (Yahoo! ID required)

<*> To change settings via email:
    mailto:hercules-390-digest@...
    mailto:hercules-390-fullfeatured@...

<*> To unsubscribe from this group, send an email to:
    hercules-390-unsubscribe@...

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/

[Alexey Bozrikov]

Fish,

I've  been  reading  your  README.AIX, as intended to be supplied with
Hercules  3.07  and noticed the last section about networking. I had a
look  at  html/herctcp.html file in the source tree, trying to realise
how  AIX  'Virtual  Ethernet'  concept  can  fit  in  Hercules  TCP/IP
networking ideology.

First  of  all,  Virtual  Ethernet only exists on pSeries and System p
machines  with  virtualisation  (allowing to communicate between LPARs
and  sharing  physical Ethernet ports between multiple virtual servers
through  VIOS).  Therefore  it  cannot  be  used  on older stand-alone
servers.  Second,  'Virtual  Ethernet'  feature  does  not  create any
character  devices  in  /dev, therefore adopting same approach as with
TUN/TAP driver is hardly possible.

TUN/TAP  driver  is not ported to AIX yet, while technically it should
be  possible, of course. I never tried to write device drivers for AIX
(last  time I did write any drivers was ages ago for SCO Xenix and SCO
UNIX).  I  will have a look at a IBM manuals regarding device drivers;
there  was  one, at least I have seen it for AIX 4.1 back in 1999. Or,
possibly, I am overestimating the complexity of the issue and there is
a more straightforward and easy solution?


Alexey


** Experiments should be reproducible - they should all fail in the same way.



------------------------------------

Community email addresses:
  Post message: hercules-390@...
  Subscribe:    hercules-390-subscribe@...
  Unsubscribe:  hercules-390-unsubscribe@...
  List owner:   hercules-390-owner@...

Files and archives at:
  http://groups.yahoo.com/group/hercules-390

Get the latest version of Hercules from:
  http://www.hercules-390.org

Yahoo! Groups Links

<*> To visit your group on the web, go to:
    http://groups.yahoo.com/group/hercules-390/

<*> Your email settings:
    Individual Email | Traditional

<*> To change settings online go to:
    http://groups.yahoo.com/group/hercules-390/join
    (Yahoo! ID required)

<*> To change settings via email:
    mailto:hercules-390-digest@...
    mailto:hercules-390-fullfeatured@...

<*> To unsubscribe from this group, send an email to:
    hercules-390-unsubscribe@...

<*> Your use of Yahoo! Groups is subject to:
    http://docs.yahoo.com/info/terms/