pnd_run.sh: move pre/post-run scripts where env is set up
Note: pre was run before env was set up and post still with env set,
now both are ran with env (PATH and LD_LIBRARY_PATH) set, so things may
break. This is unlikely though as almost nobody is using pre/post scripts.
Vars like PATH and LD_LIBRARY_PATH may cause tools used by pnd_run.sh
to misbehave or not start at all (like with GINGE as it carries ancient
GP2X libs), so make sure things from PND are not used.
Thanks! Both of these are essentially patches from j-r; the tracker urls are..
http://bugs.openpandora.org/index.php?do=details&task_id=321&action=takeownership&ids=321#
http://bugs.openpandora.org/index.php?do=details&task_id=322
pnd_make.sh: use root uid/gid for squashfs, not user's
there is no point using uid/gid of current user that's creating the
.pnd (which can be almost any uid/gid as it's on the host system),
the SD card/appdata is owned by root by default, so archived
files should be too.
It's also the reason why it wasn't possible to update existing files
(appdata and pnd uid/git did not match), but also right permissions
must be set.
skeezix [Fri, 18 May 2012 01:56:15 +0000 (21:56 -0400)]
mmenu: introduced artificial delay after SD event (ins/eject); mmenu is racing against pndnotifyd
and if mmenu wakes up first, then the lockfile is not yet present and mmenu sees minimal changes..
so if we make mmenu wait a second or two then almost certainly pndnotifyd should have awoken and
set up its lockfile. (Another option is to tweak pndnotifyd to wake up more often, but I prefer
not to bother that application as it is omnipresent .. lets avoid introducing cpu leach.)
skeezix [Fri, 18 May 2012 01:30:57 +0000 (21:30 -0400)]
mmenu: separate /pandora/mmenu scan from pnd-scan per se .. so if .desktop is scan, and .pnd is not, but an aux-pnd searchpath is defined
then attempt to pnd-disco there (regardless of .pnd scan being disabled otherwise.) This will restore expected functionality from previous
(pre-.desktop scan) releases, and generalyl will not bother anyone (few use 'mmenu' specific apps anyway.) For people who do use it, it'll
start working again, though not quite as fast as a pure .desktop scan (since there are no .desktop's for the mmenu-specific pnds, by definition.)
Nils Kneuper [Thu, 23 Feb 2012 16:20:56 +0000 (17:20 +0100)]
Changed the osversion to only accept nonNegativeInteger (in the schema).
This make checking via scripts that allow "equal" or "less than"
comparisions more easy (or possible at all...).
skeezix [Thu, 12 Jan 2012 21:53:22 +0000 (16:53 -0500)]
Preliminary stab at app scan locking; pndnotifyd creates logfile /tmp/whatever and minimenu watches out for it
mmenu will ignore 'old' lockfile (2 minutes or more), and for recent lockfile (<2mins) it will check every half second
for up to 20 times (10 seconds)
mmenu is configured in mmenu.conf so these settings can be fiddled with
skeezix [Thu, 12 Jan 2012 04:21:40 +0000 (23:21 -0500)]
Attempting fix for latest mmenu beta (using .desktop loading) not exec'ing some pnd applications.
I think -s was botched, and some attributes not used (cpu speed), which have been fixed/added now,
using additional X-Pandora attributes to avoid 'guesswork'
Nils Kneuper [Thu, 22 Dec 2011 11:00:32 +0000 (12:00 +0100)]
improve text about examples since the old wiki page skeezix created is
*not* the best source to get examples (at least when looking at all of
the page...)
skeezix [Tue, 22 Nov 2011 22:15:39 +0000 (17:15 -0500)]
Added new mmenu.conf option, to have mmenu load all visible icons as needed; combine with load defered means icons will load in background, but visible ones loaded before render
otherwise kernel opens pnds as writable and pndnotifyd gets
unwanted IN_CLOSE_WRITE event on exit, which causes a rescan and
.desktop regeneration (slow thing).
Not to mention those filesystems are readonly anyway.
skeezix [Wed, 9 Nov 2011 21:26:04 +0000 (16:26 -0500)]
mmenu; fixed .desktop and pnd scans differing apps found; .desktop was missing /usr/pandora/mmenu for mmenu specific firmware built in apps; also, was
accidentally including bad categories, where pnd-scan filters them out. As of now, app listings between .desktop and pnd-scan should be more or less
the same, unless purposefully different through various nefarious means
skeezix [Wed, 9 Nov 2011 19:44:43 +0000 (14:44 -0500)]
Can now remember last app run, so on return to mmenu it jumps right back there
--> defaults to _yes_, but used to be _no_, so people need to reset conf in the menu
or toggle it to Yes; a reflash or new users will default to yes
skeezix [Fri, 4 Nov 2011 18:36:59 +0000 (14:36 -0400)]
mmenu: background loading icons works nicely .. pretty fast, doesn't impact user speed much
conf: setting bg icon loading to default on in the deployment mmenu.conf
conf: did not yet set default to loading .desktops, still using .pnd for mmenu discovery .. overall a net speedup due to bg icon load, but even faster with .desktop
skeezix [Mon, 31 Oct 2011 15:20:49 +0000 (11:20 -0400)]
excerpt of a rant to me by Craig:
"No, you and your Lintard friends destroyed everything the Pandora could
have been, as you well know there were complex plans to make it as
revolutionary in function as it was in design, but you lintarded it up to
be a pile of shit."
---> ahh, so good to work for free under this sort of leadership ;)
Some editors remove them automatically, and people start sending patches
with such changes mixed in the patch with actual changes, which makes
those patches hard to review, so let's just sed out those spaces.
My machine in some real rare cases (once a week or so) just stops
charging without any good reason. What helps here is replugging
the charger or writing to a chip charge register, so we do the later
indirectly every 20 minutes as a workaround.
There have been similar reports in the forums, where user leaves
the machine charging overnight and finds flat battery in the morning,
so hopefully this will help them too.
Battery monitor thread now will also do charger control using
the new on/off switch provided by kernel:
- turn on charging when capacity goes below certain configured threshold
- turn it off when it reaches configured capacity AND charge current
goes below configured threshold.
skeezix [Tue, 7 Jun 2011 16:02:20 +0000 (12:02 -0400)]
Bloody off-by-1 error for a malloc, leading to ever-so-slight memory corruption. (caused 'arg' argument to randomly show up in 'info links' .desktop's)
Nils Kneuper [Wed, 9 Mar 2011 23:28:59 +0000 (00:28 +0100)]
new genpxml.sh from sebt3 with slight modifications
this script does not generate pxml files that can be validated with the
schema (yet), but at least it *does* generate some sane stuff...
skeezix [Wed, 9 Mar 2011 16:54:32 +0000 (11:54 -0500)]
libpnd; header and body for get_description() were a little out of sync; header was correct, so added wrapper function to body so as to match header, without removing the inconsistent existing function (on the off chance some app is using it.) We can deprecate it later if we like.
Nils Kneuper [Tue, 8 Mar 2011 13:41:02 +0000 (14:41 +0100)]
updated pnd_make.sh based on comments from the ML and irc:
* use genpxml.sh and the schema from the dir in which pnd_make.sh is if
* no param is supplied
* check if a PXML file was actually created during the genpxml step
* add some checks for required programs (iso and squashfs stuff)
* add some more sanity checks including aborts if an error was detected
* add missing help info for the param --genpxml
* add comment at the top of the file mentioning what the file is about
* fix various typos
Nils Kneuper [Sun, 6 Mar 2011 01:05:59 +0000 (02:05 +0100)]
new pnd_make.sh with many changes (but still compatible when usign the
old commands). Changelog:
* Add decent argument parsing.
* Add support for schema validation.
* Add support for additional switches:
--genpxml define the script to be used to generate a pxml file
--schema define the schema file to be used
* Add colored output.
* General cleanup.
Nils Kneuper [Sat, 5 Mar 2011 19:35:09 +0000 (20:35 +0100)]
New schema as discussed on the firmware-dev ml, cf this thread:
http://openpandora.org/pipermail/firmware-dev/2011-March/000056.html
Added updated examples