Mailing List Archive

Binary Distributions
I have taken the liberty to create a set of directories on hyperreal.

/export/pub/httpd/dist/binaries

Under that there is currently sunos_4.1.3, netbsd_1.1 and bsdi_1.1.
In the sunos_4.1.3 directory I have placed modules.c and Makefile.
modules.c should be portable between systems, but I leave it there
just in case.

I would like to start some discussion about what the binary releases
should include, and who is responsible for what platform. In theory,
we should just be able to grab a copy of these two files from the
directories above and compile the release.

IMHO - A binary release should simply be the source release with
compiled binaries in the 'src' directory. This way, if they do make
a change, they should not compile everything before relinking. In
theory, we have known working compiler configs etc....

The directories I have created are for the OS's I have access to
and am willing to supply binaries. I could also add hpux_9.0.x
if that is not covered by some other poor soul.
Re: Binary Distributions [ In reply to ]
On Sun, 29 Oct 1995, Randy Terbush wrote:
> I have taken the liberty to create a set of directories on hyperreal.
>
> /export/pub/httpd/dist/binaries
>
> Under that there is currently sunos_4.1.3, netbsd_1.1 and bsdi_1.1.
> In the sunos_4.1.3 directory I have placed modules.c and Makefile.
> modules.c should be portable between systems, but I leave it there
> just in case.

Okay, I added BSDI 2.0, Solaris 2.4, and Irix 5.3. Those are the
platforms I am willing and able to build for. The ownerships on those
directories can be a clue as to who is responsible for each binary build.

> I would like to start some discussion about what the binary releases
> should include, and who is responsible for what platform. In theory,
> we should just be able to grab a copy of these two files from the
> directories above and compile the release.
>
> IMHO - A binary release should simply be the source release with
> compiled binaries in the 'src' directory. This way, if they do make
> a change, they should not compile everything before relinking. In
> theory, we have known working compiler configs etc....

Keeping the .o files around will be a pain - they don't
compress very well. I would support having everything the same as the
regular distribution, except with a

> The directories I have created are for the OS's I have access to
> and am willing to supply binaries. I could also add hpux_9.0.x
> if that is not covered by some other poor soul.

no one yet, but rob hartill hasn't signed up for anything yet, and I know
it's his favorite OS.

Brian

p.s. - w/r/t the whole performance issue, www.organic.com (which is also
a virtual server for most of our clients) is a 486/50 which yesterday
handled 300444 hits without breaking a sweat.... that Sparc 20 purchase
just keeps getting pushed back further and further in priority :)

--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--
brian@organic.com brian@hyperreal.com http://www.[hyperreal,organic].com/
Re: Binary Distributions [ In reply to ]
> > The directories I have created are for the OS's I have access to
> > and am willing to supply binaries. I could also add hpux_9.0.x
> > if that is not covered by some other poor soul.
>
> no one yet, but rob hartill hasn't signed up for anything yet, and I know
> it's his favorite OS.

I said I'd do HPsUX and NeXT.

My NeXT fanatic office mate will help me produce a "fat" binary for
NeXTstep that should run on different hardware.


rob
Re: Binary Distributions [ In reply to ]
At 10:39 AM 11/14/95 -0800, you wrote:
>On Sun, 29 Oct 1995, Randy Terbush wrote:
>> I have taken the liberty to create a set of directories on hyperreal.
>>
>> /export/pub/httpd/dist/binaries
>>
>> Under that there is currently sunos_4.1.3, netbsd_1.1 and bsdi_1.1.
>> In the sunos_4.1.3 directory I have placed modules.c and Makefile.
>> modules.c should be portable between systems, but I leave it there
>> just in case.
>
>Okay, I added BSDI 2.0, Solaris 2.4, and Irix 5.3. Those are the
>platforms I am willing and able to build for. The ownerships on those
>directories can be a clue as to who is responsible for each binary build.

I already have 0.8.16 binaries for Linux_aout, and Linux_Elf on
ftp.qosina.com. I am still looking for SCO binaries, there are alot
of SCO people out there without a development system, and right now
I don't have access to SCO.
>
>> I would like to start some discussion about what the binary releases
>> should include, and who is responsible for what platform. In theory,
>> we should just be able to grab a copy of these two files from the
>> directories above and compile the release.
>>
>> IMHO - A binary release should simply be the source release with
>> compiled binaries in the 'src' directory. This way, if they do make
>> a change, they should not compile everything before relinking. In
>> theory, we have known working compiler configs etc....
>
>Keeping the .o files around will be a pain - they don't
>compress very well. I would support having everything the same as the
>regular distribution, except with a

I have to agree with Brian on this one. If they had a development system
they would compile it for themselves, us including all the .o and such
files doesn't really make a difference.

>p.s. - w/r/t the whole performance issue, www.organic.com (which is also
>a virtual server for most of our clients) is a 486/50 which yesterday
>handled 300444 hits without breaking a sweat.... that Sparc 20 purchase
>just keeps getting pushed back further and further in priority :)

We have a 386DX-40 running Linux with 16 Mb of ram, getting about
16,000 hits per day ( about 5000 during 12noon-1pm) running apache.
It's been up and running without a problem for 49 days. (Running 0.8.15)

<Aram>
--
Aram W. Mirzadeh, MIS Manager, Qosina Corporation
http://www.qosina.com/~awm/, awm@qosina.com
Apache httpd server team http://www.apache.org
Re: Binary Distributions [ In reply to ]
>
> At 10:39 AM 11/14/95 -0800, you wrote:
> >On Sun, 29 Oct 1995, Randy Terbush wrote:
> >> I have taken the liberty to create a set of directories on hyperreal.
> >>
> >> /export/pub/httpd/dist/binaries
> >>
> >> Under that there is currently sunos_4.1.3, netbsd_1.1 and bsdi_1.1.
> >> In the sunos_4.1.3 directory I have placed modules.c and Makefile.
> >> modules.c should be portable between systems, but I leave it there
> >> just in case.
> >
> >Okay, I added BSDI 2.0, Solaris 2.4, and Irix 5.3. Those are the
> >platforms I am willing and able to build for. The ownerships on those
> >directories can be a clue as to who is responsible for each binary build.
>
> I already have 0.8.16 binaries for Linux_aout, and Linux_Elf on
> ftp.qosina.com. I am still looking for SCO binaries, there are alot
> of SCO people out there without a development system, and right now
> I don't have access to SCO.

I will provide SCO binaries. I didn't realise there was an urgent requirement,
and I'm a little snowed under right now.

> >
> >> I would like to start some discussion about what the binary releases
> >> should include, and who is responsible for what platform. In theory,
> >> we should just be able to grab a copy of these two files from the
> >> directories above and compile the release.
> >>
> >> IMHO - A binary release should simply be the source release with
> >> compiled binaries in the 'src' directory. This way, if they do make
> >> a change, they should not compile everything before relinking. In
> >> theory, we have known working compiler configs etc....
> >
> >Keeping the .o files around will be a pain - they don't
> >compress very well. I would support having everything the same as the
> >regular distribution, except with a
>
> I have to agree with Brian on this one. If they had a development system
> they would compile it for themselves, us including all the .o and such
> files doesn't really make a difference.
>
> >p.s. - w/r/t the whole performance issue, www.organic.com (which is also
> >a virtual server for most of our clients) is a 486/50 which yesterday
> >handled 300444 hits without breaking a sweat.... that Sparc 20 purchase
> >just keeps getting pushed back further and further in priority :)
>
> We have a 386DX-40 running Linux with 16 Mb of ram, getting about
> 16,000 hits per day ( about 5000 during 12noon-1pm) running apache.
> It's been up and running without a problem for 49 days. (Running 0.8.15)
>
> <Aram>
> --
> Aram W. Mirzadeh, MIS Manager, Qosina Corporation
> http://www.qosina.com/~awm/, awm@qosina.com
> Apache httpd server team http://www.apache.org
>
>

--
Ben Laurie Phone: +44 (181) 994 6435
Freelance Consultant Fax: +44 (181) 994 6472
and Technical Director Email: ben@algroup.co.uk
A.L. Digital Ltd,
London, England.
Re: Binary Distributions [ In reply to ]
Re: Binary Distributions [ In reply to ]
This conversation has been formalized into a document on the internal
page:

http://hyperreal.com/httpd/binaries.html

Let me know if you have any problems, or would like to add a platform for
testing.

I would hope that after tuesday night's build, we could prepare binaries
from 0.8.17 for 1.0.

Brian

--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--
brian@organic.com brian@hyperreal.com http://www.[hyperreal,organic].com/
Re: Binary Distributions [ In reply to ]
> I would hope that after tuesday night's build, we could prepare binaries
> from 0.8.17 for 1.0.

Brian? Can you add me to the list for the OS's:

FreeBSD 2.0.5
FreeBSD 2.1 (released sometime end of this month)

> Brian

Ay.
Re: Binary Distributions [ In reply to ]
> > I would hope that after tuesday night's build, we could prepare binaries
> > from 0.8.17 for 1.0.
>
> Brian? Can you add me to the list for the OS's:
>
> FreeBSD 2.0.5
> FreeBSD 2.1 (released sometime end of this month)

Interesting.. Are these two *not* binary compatibile?
Re: Binary Distributions [ In reply to ]
FreeBSD 2.0.5/2.1 certainly are binary compatible. Sorry.
Wanna say I'm able to provide binaries for 2.+ instead? FreeBSD is
also (99.9%) compatible with BSDi, but I've seen instances of BSDi
binaries misbehaving on FreeBSD.

[.Netscape's User Agent field is made to contain the string resulting
from a call to 'uname'. The call fails on FreeBSD when running
BSDi binaries. Take a look at yer referer_log sometime]

Ay.
Re: Binary Distributions [ In reply to ]