806 - milestone branch reminder -> User agent implementation needs update

Status: VERIFIED FIXED

(SeaMonkey :: UI Design, defect, P1)

Description

[kipp]

From: [e-mail] (Christopher Yeh)

As you go about designing Raptor, would you please take a moment and decide
a simple XP way of specifing the version and user agent string?  I can't
tell you how painful it is everytime we decide to bump either in the
current 5.0 code base. You have to muck with each FE in a completely
different way.

If there was a single file that was used to config all platforms, that
would be my dream.

Thanks.

Chris Yeh

Comment 1

[kipp]

And rickg things it would be good to modify the user-agent data from our viewers
command line...

Comment 2

[Chris Yeh]

bumping up the priority, since we're now doing a raptor based 5.0. this stuff is
important for releasing betas and doing day to day qa.

Comment 3

[leger]

Putting on ss: radar

Comment 4

[leger]

Not an ss: bug.  Taking off list.

Comment 5

[don]

Re-assigned to [e-mail].

Kipp, Who should get this bug?  It doesn't appear to be an xpapps issue ...

Comment 6

[Chris Yeh]

Hubie was working on a XP single file that you use to set all of the
configuration stuff. This falls under that, so re-assigning to hshaw

Comment 7

[leger]

Setting all current Open/Normal to M4.

Comment 8

[Paul MacQuiddy]

per leger, assigning QA contacts to all open bugs without QA contacts according
to list at http://bugzilla.mozilla.org/describecomponents.cgi?product=Browser

Comment 9

[leger]

What is the satus of this?  All straightened out now for apprunner?

Comment 10

[Suresh Duddi (gone)]

Don already has a bug on this one.

Comment 11

[don]

Re-assigned to [e-mail], changed summary, target milestone to M4,
platform and OS to All, and added lots o' folks to the cc list.

Rick, please help me figure out the scope of this feature.  I need to know how
long it's going to take to implement and which engineers from what teams would
be best at writing the platform-specific work.  Thanks.

Comment 12

[don]

*** Bug 1862 has been marked as a duplicate of this bug. ***

Comment 13

[Hubert Shaw]

There currently exists a way to set the user-agent in
one file.  It does require a recompile.

This is located in nsNetlibService::nsNetlibService().
To make Chris's life easier it'd be nice if we used
some external file so no recompile is necessary, though
for now, having the ability to set the user-agent in
one place should satisfy Chris's main need.

Comment 14

[Chris Yeh]

The UA is set in nsINetService.cpp, correct? If that's the case, I (or someone on
my team) can go verify that it works alright.

Comment 15

[Hubert Shaw]

It is being initialized in:
  mozilla/network/module/nsNetService.cpp:nsNetService::nsNetService()

The output looks something like:
  Mozilla 5.0 [en] (Win95; I)

I've been informed that this bug has been
assigned to Rick to 'enhance' the contents
of the User-Agent.  I do not know the extent
of this and will defer to Rick.

Comment 16

[don]

Changed target milestone to M5.

Comment 17

[don]

Changed milestone to M6.

Comment 18

[gelderen]

*** Bug 6054 has been marked as a duplicate of this bug. ***

Comment 19

[gelderen]

Added myself to CC because of duplicate bug 6054.

I'm not sure what to do with this: if we want to set in at compile-time we can
simply add some LINUX, __FreeBSD__, __NetBSD__ macro's to the current code. If
however we want to extract the value at run-time we need platform specific code.

I can deliver patches for the first choice, and patches for runtime-detection of
the LINUX, FreeBSD and NetBSD versions.

Rick: What's the plan? Macro's for now and structural improvements later on? Let
me know and I might be able to provide some diffs.

Comment 20

[Chris Yeh]

More user agent requirements.

Steve Elmer would like to be able to identify browsers that come from MCD/CCK as
opposed to the ones available from general download. For the 4.0X series browser,
this was done by reading in values from an external netscape.cfg file at browser
initialization time that would tweak and customize the user agent field.

When the user agent gets implemented, some on-the-fly override/customization of
the user agent has to be possible. This should be in an external file so to
minimize recompiles.

Don Melton tells me the user agent won't be implemented in apprunnner until M8
(dependent on Necko) and that the apprunner initialization code that reads
external preferences files won't be ready until M10 or so.

Comment 21

[selmer (gone)]

In addition to the "append" user agent string mechanism using the netscape.cfg
and preferences (described above by cyeh), we also need to have a tag in the
string that identifies the _Netscape_ distribution channel as MCD/CCK and isn't
modifiable once the bits are in the can.  I don't yet know if MCD and CCK need
separate tags or if a single tag will work for both.

So, there's a tag we need that's burned into the bits and another we need that
can be modified at run time using the prefs.  (I believe this is the same
requirement we had in 4.5.)

Comment 22

[selmer (gone)]

In addition to the "append" user agent string mechanism using the netscape.cfg
and preferences (described above by cyeh), we also need to have a tag in the
string that identifies the _Netscape_ distribution channel as MCD/CCK and isn't
modifiable once the bits are in the can.  I don't yet know if MCD and CCK need
separate tags or if a single tag will work for both.

So, there's a tag we need that's burned into the bits and another we need that
can be modified at run time using the prefs.  (I believe this is the same
requirement we had in 4.5.)

Comment 23

[selmer (gone)]

OK, I had a misconception.  It appears that both kinds of tagging are driven by
the same pref in 4.5.  If the pref exists, then the user agent string gets the
"C-" stuff inserted and then the value of the pref gets added as well.

Comment 24

[selmer (gone)]

*** Bug 6117 has been marked as a duplicate of this bug. ***

Comment 25

[don]

Re-assigning to me for M10 (although we may get it done for M9).

Comment 26

[don]

Moved to M8.

Comment 27

[don]

Resolved as FIXED since the other bugs are resolved.

Comment 28

[Christian Wenz]

on Milestone 14, naviator.userAgent returns this:
Mozilla/5.0 ([os-dependen information]) Mozilla/m13
imho, it should be ....Mozilla/m14
You can see this effect under Help->About

Comment 29

[chris hofmann]

over to leaf.  putting on check list of things that need to be done
on the milestone branches from here on out...

Comment 30

[chris hofmann]

changing the title

Comment 31

[selmer (gone)]

We really should not have recycled this bug from a [FEATURE] to a build 
problem...

Comment 32

[Phil Peterson]

PDT- This is already in hand with another bug.

Comment 33

[Daniel (Leaf) Nunes]

phil, which bug has this bug in hand?

Comment 34

[Daniel (Leaf) Nunes]

alright, this is something that gets handled each milestone. closing bug.

Comment 35

[Blake Ross]

verified