What is the "gnu" hosttype installing gnu-ware?

What is the "gnu" hosttype installing gnu-ware?

Post by Shing-Fat Fred M » Mon, 13 Jan 2003 12:04:50



Hi,

I'm in the midst of testing out a newly built gcc before
installing it.  To test it, I have to install dejagnu.  I have
to provide the host type to the config script.  To find
a valid host type, I supply my best guess of a host type
to the config.sub script...but it doesn't recognize any
of my guesses.  Based on "uname -a":

   SunOS malawi 5.8 Generic_108528-15 sun4u sparc SUNW,Sun-Blade-100

I guessed the following

    sun
    sun5.8
    solaris
    solaris8
    blade
    blade100

I looked in one of many case statements in the
script, it takes sun*os*, though I can enter
anything in place of the asterisks.  Doesn't exactly
instll confidence that I choose the right things to
put in place of the asterisks.

What is the right way to specify the host type?
I'm installing dejagnu-1.4.3.

Thanks.

Fred

--

Carleton University, Dept. of Electronics
1125 Colonel By Drive, Ottawa, Ontario
Canada, K1S 5B6

 
 
 

What is the "gnu" hosttype installing gnu-ware?

Post by Dave Uhrin » Mon, 13 Jan 2003 15:49:43



> Hi,

> I'm in the midst of testing out a newly built gcc before
> installing it.  To test it, I have to install dejagnu.  I have
> to provide the host type to the config script.  To find
> a valid host type, I supply my best guess of a host type
> to the config.sub script...but it doesn't recognize any
> of my guesses.  Based on "uname -a":

>    SunOS malawi 5.8 Generic_108528-15 sun4u sparc SUNW,Sun-Blade-100

Ultra 1 running Solaris 9.  Just execute ./configure --help to see your
options.  It configures and builds with just

$ ./configure --prefix=/usr/sfw

But it's not going to give you much confidence in your compiler.  The
tests are not accurate and your compiler will fail at least 2 tests
unexpectedly.

 
 
 

What is the "gnu" hosttype installing gnu-ware?

Post by Shing-Fat Fred M » Tue, 14 Jan 2003 04:16:51




>>Hi,

>>I'm in the midst of testing out a newly built gcc before
>>installing it.  To test it, I have to install dejagnu.  I have
>>to provide the host type to the config script.  To find
>>a valid host type, I supply my best guess of a host type
>>to the config.sub script...but it doesn't recognize any
>>of my guesses.  Based on "uname -a":

>>   SunOS malawi 5.8 Generic_108528-15 sun4u sparc SUNW,Sun-Blade-100

>Ultra 1 running Solaris 9.  Just execute ./configure --help to see your
>options.  It configures and builds with just

>$ ./configure --prefix=/usr/sfw

>But it's not going to give you much confidence in your compiler.  The
>tests are not accurate and your compiler will fail at least 2 tests
>unexpectedly.

Thanks, I ended up doing just that (I've seen this sort
of build procedure before with other gnu-related apps).
The README for this one seems a little different, but it
might be out of date.  So much good software, and not
a volunteer to document....then again, I sort of see why,
if I ask myself whether I would ever get enough expertise
or time to volunteer myself.  I guess that's what newsgroup
archives are for.

Regarding the host type above, I still need to figure out
what the different ways to refer to my host type are.
This is because I will be comparing the failed tests with
other submitted test results, and they have ways of
referring to their machine that I find confusing.  For
example, I always thought that my OS above was solaris 8
even though it never explicitly says so.  What if I was
wrong all this time?  On the test results newsgroup,
people refer to their machine as (for example)
sparc-sun-solaris2.8....what is that?  How can we have
such a divergence in version numbers (2.8 vs. 5.8)?

Times like this, one wonders how it is that sun OS's
got such a diverse way of being named.  I suspect
that 2.8 might be for PCs, but that's in contrast with
other OS "codes" that have i86 explicitly embedded in
the names.

Fred

--

Carleton University, Dept. of Electronics
1125 Colonel By Drive, Ottawa, Ontario
Canada, K1S 5B6

 
 
 

What is the "gnu" hosttype installing gnu-ware?

Post by Dave Uhrin » Tue, 14 Jan 2003 06:42:57





>>>Hi,

>>>I'm in the midst of testing out a newly built gcc before
>>>installing it.  To test it, I have to install dejagnu.  I have
>>>to provide the host type to the config script.  To find
>>>a valid host type, I supply my best guess of a host type
>>>to the config.sub script...but it doesn't recognize any
>>>of my guesses.  Based on "uname -a":

>>>   SunOS malawi 5.8 Generic_108528-15 sun4u sparc SUNW,Sun-Blade-100

>>Ultra 1 running Solaris 9.  Just execute ./configure --help to see your
>>options.  It configures and builds with just

>>$ ./configure --prefix=/usr/sfw

>>But it's not going to give you much confidence in your compiler.  The
>>tests are not accurate and your compiler will fail at least 2 tests
>>unexpectedly.

> Regarding the host type above, I still need to figure out
> what the different ways to refer to my host type are.
> This is because I will be comparing the failed tests with
> other submitted test results, and they have ways of
> referring to their machine that I find confusing.  For
> example, I always thought that my OS above was solaris 8
> even though it never explicitly says so.  What if I was
> wrong all this time?  On the test results newsgroup,
> people refer to their machine as (for example)
> sparc-sun-solaris2.8....what is that?  How can we have
> such a divergence in version numbers (2.8 vs. 5.8)?

Perhaps because the people who write those configure scripts don't know
that the 2.X designation is no longer used, or maybe that keeps things in
order since the earlier releases started with Solaris-2.  The OS itself is
5.X; see below.

Quote:> Times like this, one wonders how it is that sun OS's
> got such a diverse way of being named.  I suspect
> that 2.8 might be for PCs, but that's in contrast with
> other OS "codes" that have i86 explicitly embedded in
> the names.

Nope, the GNU way of describing Solaris 9 on SPARC is either 2.9 or 5.9.
Other GNU software describes this machine the same way you mentioned.
Presumably you could use the following to describe the system.  From the
top of config.log in the dejagnu source tree:

## ---------- ##
## Platform.  ##
## ---------- ##

hostname = tarfu
uname -m = sun4u
uname -r = 5.9
uname -s = SunOS
uname -v = Generic_112233-03

/usr/bin/uname -p = sparc
/bin/uname -X     = System = SunOS
Node = tarfu
Release = 5.9
KernelID = Generic_112233-03
Machine = sun4u
BusType = <unknown>
Serial = <unknown>
Users = <unknown>
OEM# = 0
Origin# = 1
NumCPU = 1
uname -v = Generic_112233-03

/usr/bin/uname -p = sparc
/bin/uname -X     = System = SunOS
Node = tarfu
Release = 5.9
KernelID = Generic_112233-03
Machine = sun4u
BusType = <unknown>
Serial = <unknown>
Users = <unknown>
OEM# = 0
Origin# = 1
NumCPU = 1

/bin/arch              = sun4
/usr/bin/arch -k       = sun4u

 
 
 

What is the "gnu" hosttype installing gnu-ware?

Post by Shing-Fat Fred M » Tue, 14 Jan 2003 13:03:43


Quote:

>>Regarding the host type above, I still need to figure out
>>what the different ways to refer to my host type are.
>>This is because I will be comparing the failed tests with
>>other submitted test results, and they have ways of
>>referring to their machine that I find confusing.  For
>>example, I always thought that my OS above was solaris 8
>>even though it never explicitly says so.  What if I was
>>wrong all this time?  On the test results newsgroup,
>>people refer to their machine as (for example)
>>sparc-sun-solaris2.8....what is that?  How can we have
>>such a divergence in version numbers (2.8 vs. 5.8)?

>Perhaps because the people who write those configure scripts don't know
>that the 2.X designation is no longer used, or maybe that keeps things in
>order since the earlier releases started with Solaris-2.  The OS itself is
>5.X; see below.

>>Times like this, one wonders how it is that sun OS's
>>got such a diverse way of being named.  I suspect
>>that 2.8 might be for PCs, but that's in contrast with
>>other OS "codes" that have i86 explicitly embedded in
>>the names.

>Nope, the GNU way of describing Solaris 9 on SPARC is either 2.9 or 5.9.
>Other GNU software describes this machine the same way you mentioned.
>Presumably you could use the following to describe the system.  From the
>top of config.log in the dejagnu source tree:

>## ---------- ##
>## Platform.  ##
>## ---------- ##

>hostname = tarfu
>uname -m = sun4u
>uname -r = 5.9
>uname -s = SunOS
>uname -v = Generic_112233-03

>/usr/bin/uname -p = sparc
>/bin/uname -X     = System = SunOS
>Node = tarfu
>Release = 5.9
>KernelID = Generic_112233-03
>Machine = sun4u
>BusType = <unknown>
>Serial = <unknown>
>Users = <unknown>
>OEM# = 0
>Origin# = 1
>NumCPU = 1
>uname -v = Generic_112233-03

>/usr/bin/uname -p = sparc
>/bin/uname -X     = System = SunOS
>Node = tarfu
>Release = 5.9
>KernelID = Generic_112233-03
>Machine = sun4u
>BusType = <unknown>
>Serial = <unknown>
>Users = <unknown>
>OEM# = 0
>Origin# = 1
>NumCPU = 1

>/bin/arch              = sun4
>/usr/bin/arch -k       = sun4u

Thanks, Dave.  Stupid me, I should have
read the html page more carefully than the
plaintext one.  It is pretty clear that it's better
to let the script choose the host for native
targets.  It chose the right one, but I'll keep
the details.  Undoubtedly, there will be a
time when I have to know them again.

I got boatfuls of failed tests.  I also found
that my files on disk are getting isolated point
corruptions.  Not sure if that is the cause
of the failures. I doubt it, because it was a
C syntax error.  It would have broken the
build procedure if it was there during the
build.  (I grepped the output from
the build steps to make sure that all occurances
of "error" and "warning" looked somewhat benign).

I'm starting the build from scratch, just in case,
so I can't even say what was the general area
of the bulk of the failures.  I erased the whole
build directory.

I also visited the website where other people
submitted their errors.  Trying to fathom the
nature of the errors is not a task for the faint
of heart.  Sure, the exact status of the test is
well documented, but what exactly the test is,
who knows.  If I get the same errors, I'm going
to go ahead and use the build anyway.  Take
a snapshot of --prefix directory before installing.

Thanks again.

Fred

--

Carleton University, Dept. of Electronics
1125 Colonel By Drive, Ottawa, Ontario
Canada, K1S 5B6