Release PNDManager


Oh, and I see, that PNDManager also scans subdirectories of the PND directores (see output line above: .../desktop/alt/...). I put the PND in subdirectory "alt", because then it does not appear on the desktop, i.e. is "not installed" then. So I think PNDManager should also not consider PNDs installed, if they are in a subdirectory of desktop/, menu/, apps/...)

I'm actually surprised that you managed to remove them from the desktop in that way since subdirectories of these directories are still visible to libpnd.


This is why all my 'hidden' PNDs are out of the 'pandora' directory entirely and stuck in a folder called SDCARDNAME/Other/ where they are neatly hidden from libpnd.

Currently the location section in the Installed view only shows which card the PND is installed to. Ideally I'd like to see the full path split across three lines: media, path, filename. eg:


Media: /media/SDCARDNAME/


Location: pandora/menu/


Filename: application.pnd


In the same view I'd also like to see an indicator as to whether that PND exists in the repo. (or which repo if any and which version if different.)
Yeah, that takes quite a bit of valuable space. What would you take out to make it fit?

On my screen theres a significant area of whitespace underneath the existing information in the left panel, why would anything need to be taken out?

Currently the download resume is automated, if download cuts it will try it again 5 times resuming where it left. However, it won't work if you start the download again. (This is because libpndman uses temporary files which are removed when handle is freed) However, It's not impossible to change how this works.
Hmm..... Over what kind of time frame do the 5 resume attempts occur? Entirely possible I didn't get wifi reconnected until after the 5th attempt, but it seems (based on how PNDManager reacted) that the handle was not freed until I triggered the restart. Perhaps allow for a manual resume to be triggered at any time while the handle exists? then a button could be added to the application data panels when a 'downloading' application has no active download.


- Neelix
 
This list of PNDs to update didn't include one which I knew had been updated, when I clicked into it, it then gave me the option to update it from there. Hope that makes more sense now?(6am starts in work do not contribute to my written elegance sometimes!!)
Yep, will need to think how it could get that way :)

On my screen theres a significant area of whitespace underneath the existing information in the left panel, why would anything need to be taken out?
You mean right side? It's probably the space reserved for the description. Check a PND with a longer one.
 
On my screen theres a significant area of whitespace underneath the existing information in the left panel, why would anything need to be taken out?
You mean right side? It's probably the space reserved for the description. Check a PND with a longer one.


erm... yes... the other left! *facepalm*


Even with the description taking up all 4 lines allowed for it there is still more whitespace below that.


- Neelix
 
Hm. Will think about it. Won't promise any immediate changes, since there's a whole lot of backend stuff I want to do first. Like integrating ratings and comments.
 
Really fantastic work guys :)


One oddity I noticed.. after an upgrade, I sometimes see blinking between the progress bar and the remove/upgrade/launch buttons. (Like the design, nice colour coding and everything :)


I think if you add one thing, it'd be ready for inclusion in the firmware (space permitting) -- on first run (or a run after say a month), where the DB is not present or old, pop up a dialog or panel offering to do a DB sync; as is, you're fiddling around trying to find out wwhy theres no upgrade buttons, nothign in the app category menu, etc. (Perhaps empty categories shoudl show 'try syncing to find apps!', as an additional hint.,


The icons for triggers are sort of hard to see what they are (small) until you know what they are.


Utterly fantastic though :)


jeff


edit: ahh, pndmanage got very confused when the SD card got full ;) Some error handling there could help, but yeah, not as big a case :)
 
Last edited by a moderator:
Hmm..... Over what kind of time frame do the 5 resume attempts occur? Entirely possible I didn't get wifi reconnected until after the 5th attempt, but it seems (based on how PNDManager reacted) that the handle was not freed until I triggered the restart. Perhaps allow for a manual resume to be triggered at any time while the handle exists? then a button could be added to the application data panels when a 'downloading' application has no active download.

Don't worry about this anymore, you can now resume download even if pndmanager crashes.


The only time resume doesn't work is if pndmanager cleanly shutdowns (removes all temporary files) or the download completes.


There was also bug, well err.. feature that the statistics restart even when you actually resumed (progressbar starts from zero).

I have a Mac, and put PNDs onto the card using the Mac most of the time, and Mac OS X creates those "._" files for each file it puts onto the card. Very annoying, but - afaik - not avoidable.


Maybe PNDManager should skip all files beginning with "." (hidden files) in general?

Enabled now by default
 
Last edited by a moderator:
Upgrade all seems flaky..


Hit that, and then it seemed to toggle between showing two items in the downloading bar and all of them in the bar.. eventually it downloaded 2 things, and then left the rest in upgradable section again.


Seems not harmful, but didn't work as expected :)


jeff


Does it queue things up? If thered 10 things in the Downloading.. is it trying them all at once, or just the top 1/2/3 items? (If its doing all of them.. ugggh :)
 
Last edited by a moderator:
I think my earlier issues with PND Manager may not turn out to be anything to do with the software itself, I think I have that issue discussed elsewhere of my wifi connection being very unreliable under the new kernel? Tonight I tried using is with my wireless connection and it kept crashing out, under a wired USB ether connection it works perfectly!


(null)
 
I just had PNDManager crash on sync again.


It seems to happen anytime I try to sync after the kernel (3.2 under SZb5a) has quietly stopped routing TCP/IP packets correctly. (I tried to ping my other machine immediately after the crash and got the host unreachable error.)


Here's the pndrun_pndmanger.out file.


- Neelix
 
Looks like a bug in the code somewhere.. a double free(), or not checking the return value from something..


Fun fun .. :)


jeff
 
Yay, feedback :) . This is the kind of stuff that happens when you change a lot of stuff at once, which is why I went for the public beta route. I'll look into the above stuff before starting on any new features (except maybe a couple of really small ones ;) ). Bugs first, features last is a good development order in my opinion :) .
 
That's kinda hard to replicate, given the extraordinary circumstances. Cloudef, any idea what could segfault in pndman? I can't really see what could go wrong on pndmanagement-plugin's side at that time since I just react to the data from pndman. Will check more carefully, of course.


Also, nice that the resume worked, though :)
 
Last edited by a moderator:
That's kinda hard to replicate, given the extraordinary circumstances. Cloudef, any idea what could segfault in pndman? I can't really see what could go wrong on pndmanagement-plugin's side at that time since I just react to the data from pndman. Will check more carefully, of course.

Seems to be somewhere in the curl backend, gdb backtrace would be helpful.


B-ZaR, I suggest setting pndman_set_color(0); since it makes pnd.out look not fun.


(or better, log the messages to seperate log file yourself with the hooks.)


Also, I'm suprised there are still invalid XML's floating around... I might need to make my ugly stripping code more effective, would be better if authors fixed their pxml's though. (first pnd seems to use & which should be &amp, and other one has unquoted variable...), I might be able to workaround the first one but not the second.


E: both pnd's seems to crawl fine here, with the current stripping code.. weird.
 
Last edited by a moderator:
I'm just optimizing the PXMLs of my PNDs.


Is there some tool available, which makes this easier? There is so much redundant info in the PXMLs (3x the same description, 2x the same version number - so much copy and paste!), some good tool would probably help a lot of developers, be it because they are just lazy, or in general to prevent errors due to forgotten field updates...
 
@hmc the information is only redundant if there is 1 app inside the PND.


the package part of the PXML is all about the PND itself not necessarily the app within.


the applications part of the PXML (there can be more than one application within a PND) is about the individual apps.


so it may seem redundant but its not, and the only reason you may have to do more than 2 descriptions if there is only one app inside the pnd is if you want the description in more than one language.
 
There are tools to package pnd files. Also there are examples. You should start crawling the wiki.
 
Back
Top