There's no flag to enable hildon interface when building for lpia

This is the first debdiff for proofreading

Fixed wrong debian/rules flag

Here's a quick review:

1) the debdiff touches debian/patches/clam-mail.1.patch, doesn't seem to change the meaning of the patch; would be nice to drop this gratuitous change to ease review for the other claws-mail people which might want to review this (e.g. the Ubuntu folks with interest in claws-mail or the Debian folks if we send them the patch)

2) I wonder whether it would make sense to:
a) not use ENABLE_UME / ENABLE_HILDON high level defines but instead use ENABLE_$feature for granular features; e.g. ENABLE_LARGE_WINDOWS or ENABLE_POWER_SAVING; I think this is better style autofoo which results in less hardcoding of platforms in all the source code; this might result in more readables "#ifdef HAVE_hildon_gtk_entry_set_input_mode" style
b) define MAEMO when you define UME
c) drop the UME define altogether and instead make sure you define MAEMO but not CHINOOK

what do you think?

3) An AC_ARG_ENABLE(ume, ...) is needed before using things like ac_cv_enable_ume

4) I have to admit I don't know about multiple PKG_CHECK_MODULES(FOO, ...) calls, are these additive?!

5) The AC_DEFINE(UME, 1, Build for maemo) should be "Build for ume"

Generally looks good, but I have having us to maintain the pile of UME ifdefs just to avoid the CHINOOK ifdef and the a couple of hildon specific functions; are these used for other things? Are we missing some CHINOOK stuff? Why are we missing these functions?

Thanks for your work!

Thanks for the review.
I totally agree with your point 2 but right now I don't believe we have the time to do it so, I would like to sugest we split the difference by fixing 1, 3 and 5 for now and leave 2 and 3 (which is a bit of an ugly hack I admit) for a next release.
About the IFDEFs (which are connected to the item 2 as well), the problem is that the upstream support for maemo is very specific and things we don't support like GTK's tap-and-hold and osso_uri have to be disabled when building it for UME. This item can be addressed later when some further granilarity will be added. For now I believe is good enough in terms of funcitionality and speed as one can't beat "done".

Ok, please update the debdiff with points 1, 3, 5 fixed and push to the ppa if yesterday's binaries work for you. :)

Sorry, point 3 is bogus; I misread the name of the variable used in configure.ac for the internal name used by configure; it's fine like this.

I see you moved from --enable-ume to --enable-maemo: what are the differences between the two?

If you opt for --enable-maemo, then most of the patching can go away as we're the only who are going to use --enable-ume.

As I said, upstream support for maemo is very specific but we share many things so I could have used --enable-ume as well. It was just a matter of choice but you're right, --enable-ume makes more sense. I'll change that.

If both would work, what are the differences?

The difference is the way configure.ac works. If I use the --enable-ume method, I explicitely say that I want that platform, the other way the same flag works for maemo and ume builds but it's more confusing for the one who looks at it for the first time so I'm switching to --enable-ume. The precious was a bad choice.

https://bugs.maemo.org/show_bug.cgi?id=2080 is a good example on how to write more granular checks for the available functions.

Patch redone according to the review.

Adilson: wrong patch, this is identical to the previous clawshildonfix2.diff you attached earlier.

Sorry about the mistake. Please, check this one.

Oh wow...this will be fun...

claws-mail just upgraded to 3.3.1...

is it possible to apply this patch to it? or much better, to push that upstream?

Regarding the availability of claws-mail in universe and the lack of potential mobile devs in universe, it's being hard to maintain this diff over time.

Ok,

I'm porting the diff to 3.3.1 right now...
It would be cool, if you could write an FFe (because it's a new feature) for this...

Regards,

\sh

Ok,

lool, adilson, I ported the diff to 3.3.1.
please find attached the debdiff to 3.3.1-1ubuntu2 ...please testbuild it on lpia and test run it.

please file an FFe (add it to this bug report and subscribe motu-release) with the explanation what the diff does, what new features it adds, etc.

motu-release can assign me to this bug, so I can upload this to hardy.

Regards,

\sh

ok,

no FFe needed...I just talked to our release team...
I'll build an lpia package tonight and test it...sadly I can't test the input changes ;)

\sh

Hi. Thanks for the effort.
Your patch was not working well with the lpia build. I made a few corrections and it's ok now.

This bug was fixed in the package claws-mail - 3.3.1-1ubuntu2

---------------
claws-mail (3.3.1-1ubuntu2) hardy; urgency=low

   * Updated LPIA features (LP: #198861)
   * debian/rules
    - Added --enable-maemo flag to enable the hildon interface and
      small screen changes if claws-mail is built for lpia. (LP: #198861)
   * debian/control
     - Added lpia dependencies.
   * configure.ac
     - Added test for patched GTK supporting TAP-AND-HOLD.
     - Added test for Maemo's tablet browser interface.
   * Added IFDEFs based on the flags above to allow or not those resources
     to be used. Right now UME has neither but any change on that should be
     automatically detected and added to the build.
   * Patch provided by Adilson Oliveira <email address hidden>

 -- Stephan Hermann <email address hidden> Wed, 19 Mar 2008 23:36:10 +0100