[fightnight]

Hi, just to debug and my feedback:

After upgrading to #0529, fiq patches works very well and no more freezes. Now XBMC shows an error notification on boot about cec pulse-eight but everything works great.. just the notification on boot

Pic:



#0529 to #0531 -> raspberry doesn't output any composite video (but its working in background). Replaced same files, same result. Reverted to #0529 and its working again

Today:

#0529 to #0601 -> Composite video its working again but now i can't get my usb remote working. Reverted to #0529 and works again

Log: http://pastebin.com/X0b4j7Dp

I don't want to moan, just my feedback. Thanks for your hard work.

[Milhouse]

Can you tell if your USB remote is working with #0531, maybe hearing GUI clicks or key_down entries in a debug log?

Not sure about the CEC error, anything in a debug log?

[ijsbeer79]

Wow, 'fasthash' part is opening the series tab much faster. Amazing what the differences are with good old Frodo

[FAMMAR]

Wow, 'fasthash' part is opening the series tab much faster. Amazing what the differences are with good old Frodo

fasthash?.......what's that?

[Milhouse]

PR4817, introduced in build #0531, though I think it is meant to improve scan/scrape speed when adding new tvshow content rather than library access.

[craigbeat]

Is there any way to make requests for PRs? This one: https://github.com/xbmc/xbmc/pull/4753 looks interesting, but will probably break a few things. If I get a chance at the weekend, I might see if I can try a build with it in.

If it's a bit of a pain to do, don't worry.

[popcornmix]

PR4817, introduced in build #0531, though I think it is meant to improve scan/scrape speed when adding new tvshow content rather than library access.

Yes, speeding up scanning for new content was my understanding of the benefit.

There were a couple more PR's from the same author that I've now added to newclock3 which could help other use cases
(it will use a cheaper file existence check which could help video startup time when playing from a crowded directory).

But I wasn't expecting library view access to be faster.
@ijsbeer79 are you sure the speedup is due to fasthash?
There are a number of speedups to library views if you are coming from Frodo, but I'm not sure fasthash is responsible for this
(although would be interested to hear if latest build is noticeably faster than previous build).

[Milhouse]

If they're XBMC PRs, and seem fairly sane, and stand a good chance of being accepted eventually (ie. nothing too contentious) then yeah, I can include them, although I need to be careful as every extra PR has the potential to break the build if/when the PR rots and/or introduces changes that conflict with newclock3. This particular PR looks like it might cause a few problems but I'll try and include it in the next build and see what shakes out.

---

There were a couple more PR's from the same author that I've now added to newclock3 which could help other use cases
(it will use a cheaper file existence check which could help video startup time when playing from a crowded directory).

So it looks like PR4817 has been binned in preference to PR4837.

Edit: Having read all the comments on PR4837 it might be best to wait a few more days before including it. A ton of comments on a PR is not usually a good sign, best to wait until comments die down after reaching some sort of consensus...

PR4839 looks interesting.

[popcornmix]

Edit: Having read all the comments on PR4837 it might be best to wait a few more days before including it. A ton of comments on a PR is not usually a good sign, best to wait until comments die down after reaching some sort of consensus...

Did you really mean PR4837? (it seems uncontroversial with only one comment about a blank line...

[Milhouse]

Did you really mean PR4837? (it seems uncontroversial with only one comment about a blank line...

Yes, there are a lot of collaborator comments from jmarshallnz (not sure why you're not seeing them). Several of the comments are querying if the proposed changes will actually have any benefit and may even slow down NFS users...

[tomtomclub]

cec work on my samsung D8000 smart tv thanks i am happy

[Milhouse]

Errr weird, I just refreshed PR4837 and the long running jmarshallnz conversation (30+ comments) has now disappeared, like deleted... maybe it was linked to an outdated diff?

[Milhouse]

@popcornmix: It looks like the mkortstiege commits you added to newclock3 no longer match the diffs that are now on PR4837, you've got two extra commits (probably deleted/squashed from PR4873?) that probably need to be dropped now:

less expensive CDirectory::GetDirectory using dir flag DIR_FLAG_NO_FILE_INFO (link, and comments from jmarshallnz, which may explain why it has been dropped)
use CFileItem size and date values in case they are available while path hashing (link)

Lord knows what else has changed.

[Milhouse]

@popcornmix: PR4838 conflicts with newclock3 commit Update XCode project file.

PR4837 (not from newclock3), PR4839 (not from newclock3) apply OK, at least right now they do anyway...

@craigbeat: PR4753 no longer applies on top of master and needs to be rebased (updated). I think this PR will have to wait, sorry.

[Mafarricos]

Did some commit change something about shutil.rmtree?
Im getting errors here when trying to download subtitles.

Logfile example:
http://xbmclogs.com/show.php?id=216294

The problem is related with temp directory.

Changing something like this:
__temp__ = xbmc.translatePath(pjoin(__profile__, 'temp')).decode("utf-8")
To
__temp__ = xbmc.translatePath(pjoin(__profile__, 'temp'))

Solves the problem, but I don't understand why the error is happening now.

I'm on latest release.