[Gpe-list] [PATCH] Support common buttonmap for HH.org-maintained devices

[Paul Sokolovsky]

Hello Florian,

Thursday, December 14, 2006, 8:05:40 PM, you wrote:

> Hi Paul,

> Paul Sokolovsky schrieb:
>>       Since the version of 2.6.16-hh8, most HandHelds.org-maintained
>> PDA-design (vs XDA, phone, etc.) devices switched to use common
>> button keycode map. This allows to overcome issues we previously had
>> with application buttons not working from a device to device, power
>> button not working, etc.

> that's good news... it will make our life easier in future.
> This might be good to push upstream with a little bit of documentation about how
> to map buttons on new devices so that other porting projects starting from
> upstream kernel do it in the right way too.

  Well, so far, this is HH.org-local effort to clean up and
standardize its own ports.

  But I agree that eventually, userspace GUI frameworks should provide
guidelines for the ports how to map buttons on their kernel level (or
kernel-userspace, like loadkey), not push adhoc mappings to frameworks
themselves. And maybe later, there indeed will be de-facto standard
suitable for RFCing on mainline kernel level.

>>       The change affects following devices supported by GPE: h1910,
>> h2200, h4000, hx4700, asus716. Moreover, it allows to add support for
>> new devices more easily, and it will "just work" (new devices
>> belonging to already supported PDA families like HP iPaq and Asus
>> MyPal, will simply work out of the box).
>> 
>>       The patch has been in testing since August, and by now has
>> proven to work as expected. It is also committed to OpenEmbedded
>> (org.openembedded.dev branch).

> ok, very good


>>       So, I would like to ask it to be merged into GPE mainline. The
>> change affects xserver-common.

> I'll check this in yes - just one question: I'm not sure how the shell evaluates
> this case statement - so what happens on a H6300? I assume that matches at last
> and its special map is applied at last then, right?

  Yes, patterns are processed from beginning to end, and stop on first
match. According to Single Unix Specification (
http://www.opengroup.org/onlinepubs/009695399/utilities/xcu_chap02.html#tag_02_09_04_05
):

"
In order from the beginning to the end of the case statement, each pattern that
labels a compound-list shall be subjected to [...] expansion, []  and the result
of these expansions shall be compared against the expansion of word, according
to the rules described in Pattern Matching Notation []. After the first match,
no more patterns shall be expanded, and the compound-list shall be executed.
"

   bash manpage has the same clause. So yes, exceptions and overrides
should be listed before generic rules, and generic rules should be at
the very end.



> Greetings

> Florian


Thanks!


-- 
Best regards,
 Paul                            mailto:pmiscml at gmail.com