7:06 p.m.
The User-agent should say "RISC OS", not riscos.
Is there a way we can easily get something more accurate for the hw than
arm-acorn (e.g. machine type like "RiscPC", "Iyonix" or CPU; ideally
detect
VRPC too)?
James (on DS)
7:33 a.m.
On Sun, 2007-02-04 at 10:06 +0000, James Bursa wrote:
The User-agent should say "RISC OS", not riscos.
Is there a way we can easily get something more accurate for the hw than
arm-acorn (e.g. machine type like "RiscPC", "Iyonix" or CPU; ideally
detect
VRPC too)?
These are issues with UnixLib, and should be fixed there, rather than
adding specialisation to NetSurf, I feel.
B.
1:48 a.m.
In message <1170628427.5803.2.camel@trite>
Rob Kendrick <netsurf-dev(a)rjek.com> wrote:
On Sun, 2007-02-04 at 10:06 +0000, James Bursa wrote:
> The User-agent should say "RISC OS", not riscos.
>
> Is there a way we can easily get something more accurate for the hw than
> arm-acorn (e.g. machine type like "RiscPC", "Iyonix" or CPU;
ideally detect
> VRPC too)?
These are issues with UnixLib, and should be fixed there, rather than
adding specialisation to NetSurf, I feel.
I'm not convinced that this is a problem with UnixLib. It seems to me
that UnixLib just doesn't provide precisely what we want.
Best regards,
--
Michael Drake (tlsa)
http://www.smoothartist.com/
1:52 a.m.
On Mon, 2007-02-05 at 16:48 +0000, Michael Drake wrote:
In message <1170628427.5803.2.camel@trite>
Rob Kendrick <netsurf-dev(a)rjek.com> wrote:
> On Sun, 2007-02-04 at 10:06 +0000, James Bursa wrote:
>> The User-agent should say "RISC OS", not riscos.
>>
>> Is there a way we can easily get something more accurate for the hw than
>> arm-acorn (e.g. machine type like "RiscPC", "Iyonix" or CPU;
ideally detect
>> VRPC too)?
>
> These are issues with UnixLib, and should be fixed there, rather than
> adding specialisation to NetSurf, I feel.
I'm not convinced that this is a problem with UnixLib. It seems to me
that UnixLib just doesn't provide precisely what we want.
UnixLib provides uname(), which amongst other things, is meant to return
the name of the OS and the type of hardware. It gets this wrong - it's
not NetSurf's fault, and it's not in NetSurf where it should be fixed.
B.
8 a.m.
UnixLib provides uname(), which amongst other things, is meant to
return
the name of the OS and the type of hardware. It gets this wrong - it's
not NetSurf's fault, and it's not in NetSurf where it should be fixed.
I agree, but in practice there may be things that depend on the current UL
output or expect no spaces. Also it's still up to us to implement this either
in NetSurf or as a patch to UL.
James
11:16 a.m.
On Tue, 2007-02-06 at 23:00 +0000, James Bursa wrote:
> UnixLib provides uname(), which amongst other things, is meant
to return
> the name of the OS and the type of hardware. It gets this wrong - it's
> not NetSurf's fault, and it's not in NetSurf where it should be fixed.
I agree, but in practice there may be things that depend on the current UL
output or expect no spaces. Also it's still up to us to implement this either
in NetSurf or as a patch to UL.
I'm working on a patch to correct UnixLib's output at the moment.
B.
2:02 a.m.
On Sun, 2007-02-04 at 10:06 +0000, James Bursa wrote:
The User-agent should say "RISC OS", not riscos.
Is there a way we can easily get something more accurate for the hw than
arm-acorn (e.g. machine type like "RiscPC", "Iyonix" or CPU; ideally
detect
VRPC too)?
Additionally, I'm not sure it's wise to reveal what machine a user has,
as that's very almost as bad as revealing the version of RISC OS they're
using. Perhaps changing UnixLib to return the likes of ARMv3, ARMv4,
ARMv5, etc, as this is more what it is meant to return, as the kernel is
meant to hide the actual nature of the hardware from the software
anyway.
B.
8:05 a.m.
Additionally, I'm not sure it's wise to reveal what machine a
user has,
as that's very almost as bad as revealing the version of RISC OS they're
using. Perhaps changing UnixLib to return the likes of ARMv3, ARMv4,
ARMv5, etc, as this is more what it is meant to return, as the kernel is
meant to hide the actual nature of the hardware from the software
anyway.
Yes, I agree.
The only use of this information in the UA that I can think of is for
statistics. Maybe we should just remove it.
James
8:11 a.m.
On Tue, 6 Feb 2007, James Bursa wrote:
> Additionally, I'm not sure it's wise to reveal what
machine a user has,
> as that's very almost as bad as revealing the version of RISC OS they're
> using. Perhaps changing UnixLib to return the likes of ARMv3, ARMv4,
> ARMv5, etc, as this is more what it is meant to return, as the kernel is
> meant to hide the actual nature of the hardware from the software
> anyway.
Yes, I agree.
The only use of this information in the UA that I can think of is for
statistics. Maybe we should just remove it.
I'm all for that; avoiding extraneous garbage in the UA string can only be
a good thing. Retaining the OS name (but definitely not version) is
arguably useful, I guess.
John.
6:19 p.m.
In message <61017.82.45.217.252.1170583563.squirrel(a)semichrome.net>
"James Bursa" <james(a)netsurf-browser.org> wrote:
The User-agent should say "RISC OS", not riscos.
Is there a way we can easily get something more accurate for the hw than
arm-acorn
This should be done when the UL on the autobuilder is updated. Thanks
to joty and rjek.
Best regards,
--
Michael Drake (tlsa)
http://www.smoothartist.com/
6025
days inactive
6076
days old
netsurf-dev@netsurf-browser.org
9 comments
5 participants
participants (5)
-
James Bursa -
John-Mark Bell -
Michael Drake -
Rob Kendrick -
Rob Kendrick