On Sunday 17 May 2009 17:13:24 Andreas.Eversberg wrote:
hi holger,
these patches are not the "final" patch. some things must be discussed
before of yourse. i removed the out commented code from the patch. some
comments like "// todo ..." are not removed. when you look at the file
http://home.eversberg.eu/stage1.html, you will see the new version.
whenever there is something to complain about, i will change it, until we
have patches that all of us agree.
please ask if you don't fully understand the patch. especially for the
BSC<->application interface, i was too lazy to describe all details. you
will find mncc.c usefull to see how the interface is handled.
Sorry,
we don't get anywhere with this. How should I comment on the stage1.html? Send
you a diff against html? Currently it is way too much work to pick the good
parts by hand and leave the other stuff out, I would really welcome patches one
could just apply.
Examples:
1.) Installation of the lib...
NO! Not via an install-data-hook... If you want the lib to be installed change
it away from a noinst lib. This should honor the --prefix one is giving when
configuring... this is only leaving the includes...
And please don't mix changing the noinst to a proper library and adding new
files... They have nothing to do with each other...
2.) ... Add mncc.c in one go and not by sprinkling changes around...
3.) Adding the timer.
The change looks fine, we will need to test it on real hardware... BUT it is
depending on your rename of the timer functions introduced later... maybe it
is time for using git and git-format-patch? So yes, patch is okay, but we can
not just apply it because of its dependencies (timer rename, new trau include,
...) It would also be appreciated if you can rediff this part as the debug
statements have been added... And you have unrelated changes like the
autorelease within this patch, something I have merged to openbsc by hand
already...
And with the change of input/misdn.c you try to sneak in a mISDNuser
dependency without changing the configure.in. I already asked why is this
needed? What is the benefit?
4.) Skipped for now
5.) Okay, should be something like the new 3rd patch as 3rd) is depending on
this one. All fine... but the title of the patch does not match the content.
The renaming has nothing to do with the change of semantic of this function.
So we need one patch for renaming the methods (to not pollute the global
namespace...) + users of it and one for changing the semantic + user of it.
6.) Skipped for now
7.) Don't you miss a change to include/openbsc/gsm_data.h? Currently the bts
is passed in terms of the data (or within a struct for data), can't this work
for for you?
sorry, need to stop here right now, I see most of your work is in the NMCC
change... so for now what about trying to get 1-7 resolved and progress on
these items?
z.