To be done or not to be done ... that's here the question ;)

everything about the next not so big update

Moderator: Moderator

User avatar
Uatschitchun
Overlord
Posts: 3189
Joined: Tue Dec 06, 2005 6:55 pm
Location: Germany
Contact:

Re: To be done or not to be done ... that's here the questio

Postby Uatschitchun » Tue Feb 27, 2007 3:00 pm

arj wrote:
Uatschitchun wrote: :arrow: Does 'make clean' and 'make distclean' clean correctly? Do these targets take care of cleaning libs/libavcodec, test.o, etc.? Me was feeling they don't !?


make clean, cleans mms
make distclean, cleans mms + libs

libavcodec takes a long time to compile so it's better only to clean it when absolutely needed.


Did test the targets. Steps:
Do a clean checkout
cp -a checkout checkout.orig
cd checkout
./configure
make && make clean
cd ..
diff -r checkout.orig checkout

If clean should only clean mms, there are still some files left even for mms ...
Nur in mms-1.0.9-patch60/: config.h.
Nur in mms-1.0.9-patch60/: config.mak.
Nur in mms-1.0.9-patch60/: configure.log.
Nur in mms-1.0.9-patch60/libs/termcap: config.cache.
Nur in mms-1.0.9-patch60/libs/termcap: config.log.
Nur in mms-1.0.9-patch60/libs/termcap: config.status.
Nur in mms-1.0.9-patch60/libs/termcap: libtermcap.a.
Nur in mms-1.0.9-patch60/libs/termcap: Makefile.
Nur in mms-1.0.9-patch60/libs/termcap: termcap.o.
Nur in mms-1.0.9-patch60/libs/termcap: tparam.o.
Nur in mms-1.0.9-patch60/libs/termcap: version.o.
Nur in mms-1.0.9-patch60/libs/tinyxml: tinystr.o.
Nur in mms-1.0.9-patch60/libs/tinyxml: tinyxmlerror.o.
Nur in mms-1.0.9-patch60/libs/tinyxml: tinyxml.o.
Nur in mms-1.0.9-patch60/libs/tinyxml: tinyxmlparser.o.
Nur in mms-1.0.9-patch60/libs/tinyxml: xmltest.
Nur in mms-1.0.9-patch60/libs/tinyxml: xmltest.o.
Nur in mms-1.0.9-patch60/po: cs.mo.
Nur in mms-1.0.9-patch60/po: da.mo.
Nur in mms-1.0.9-patch60/po: de.mo.
Nur in mms-1.0.9-patch60/po: es.mo.
Nur in mms-1.0.9-patch60/po: fi.mo.
Nur in mms-1.0.9-patch60/po: fr.mo.
Nur in mms-1.0.9-patch60/po: pl.mo.
Nur in mms-1.0.9-patch60/po: pt.mo.
Nur in mms-1.0.9-patch60/po: ru.mo.
Nur in mms-1.0.9-patch60/po: sv.mo.
Nur in mms-1.0.9-patch60/: termcap-config.log.
Nur in mms-1.0.9-patch60/termio: getch2.o.
Nur in mms-1.0.9-patch60/: test.c.
Nur in mms-1.0.9-patch60/: version.h.


after that
cd checkout
make distclean
cd ..
diff -r checkout.orig checkout

Shouldn't distclean set the source completly back, if I get it right?
Nur in mms-1.0.9-patch60/: config.h.
Nur in mms-1.0.9-patch60/: config.mak.
Nur in mms-1.0.9-patch60/: configure.log.
Nur in mms-1.0.9-patch60/libs/termcap: config.cache.
Nur in mms-1.0.9-patch60/libs/termcap: config.log.
Nur in mms-1.0.9-patch60/libs/termcap: config.status.
Nur in mms-1.0.9-patch60/libs/termcap: Makefile.
Nur in mms-1.0.9-patch60/: termcap-config.log.
Nur in mms-1.0.9-patch60/: test.c.
Nur in mms-1.0.9-patch60/: version.h.


Lg
Roman

hendrik
Posts: 15
Joined: Mon Sep 11, 2006 8:05 pm

Re: To be done or not to be done ... that's here the questio

Postby hendrik » Tue Feb 27, 2007 10:09 pm

arj wrote:
acmelabs wrote:[*]much faster navigation (45sec to enter a big folder atm)[/list]


Could you refresh my memory and tell me where it happens more specifically? Maybe provide a strace log. Thanks.


It happens if you are using remote shares as media drive and even worse if these shares are connected via WLAN. At least this is what I notice on my system, can not speak for acmelabs but I guess he has the same in mind.

The problem is again the tons of lstat's that slow the system down.

A solution would be some kind of lazy loading:

1. open the directory, get the content (that's usually quite fast)
2. only do lstat's on the files that are supposed to be shown in the gui
3. render the gui
4. do the other lstat's / collect information

Another solution (maybe in combination with the above) would be to just guess the content in the first place (file extension: it's possible but not common to have directories with a .mp3 suffix) and recheck / rerender the gui later (that would also speedup the first enter into a directory with mp3/ogg files).

And third load covers lazy, that means render first with the normal folder icon, later replace it with the cover.

Of course you need threads and locking for such a solution, so it's not trivial to implement this.

User avatar
arj
Site Admin
Posts: 2316
Joined: Thu Dec 01, 2005 8:51 pm
Location: Denmark
Contact:

Re: To be done or not to be done ... that's here the questio

Postby arj » Wed Feb 28, 2007 12:59 am

arj wrote:
Uatschitchun wrote: :arrow: Could playnow=true be made default?


It was initially off because it is sort of a different philosophy for handling audio. I have since come to the conclusion (after many hours of rockbox usage :)) that it is perhaps a smarter way.

Normally one doesn't really need to care about playlist. I just want to list to my damn music, "do that and stop bothering me with all this playlist stuff I don't care about and don't need right now" :)

The idea is that one the majority of times does not use the playlist, one just plays songs. The problem with those approaches is that the playlist becomes so secondary that most people don't use them, even though they could be of very good usage.

So you sort of have 2 camps, the geeky winamp camp where everything is a playlist. The ipod camp where there is almost to concept of playlists. MMS is very close to the winamp camp (it actually support it :)). This patch sort of bridges this gap between the two camps which is very nice. It's sort of the best of two worlds. The only problem is that the destruction of the playlist could be quite disastrous. It breaks the principle of safe-exploration.

So how do we fix this situation?

- autosave the playlist when doing the play now so that one can get the playlist back (not optimal, how would one know that it did this?)
- same as 1, just display a message the first time saying that the old playlist has been saved as XXX. Better but still not optimal.
- ask (warn) the user of what is about to happen. Sucks if one doesn't really care about this. Modal windows are the devil.

Does anyone have other suggestions?


Please let me know what you guys think. I convinced myself that the option 2 was the best, but for that we need a new message (=more translations). So if this is the case then I would like to know ASAP. The 1.0.8.2 is very close and I need to send out request for translations soon to be able to get them all back in time.

User avatar
arj
Site Admin
Posts: 2316
Joined: Thu Dec 01, 2005 8:51 pm
Location: Denmark
Contact:

Re: To be done or not to be done ... that's here the questio

Postby arj » Wed Feb 28, 2007 1:22 am

Uatschitchun wrote:
arj wrote:
Uatschitchun wrote: :arrow: Does 'make clean' and 'make distclean' clean correctly? Do these targets take care of cleaning libs/libavcodec, test.o, etc.? Me was feeling they don't !?


make clean, cleans mms
make distclean, cleans mms + libs

libavcodec takes a long time to compile so it's better only to clean it when absolutely needed.


Did test the targets. Steps:
Do a clean checkout
cp -a checkout checkout.orig
cd checkout
./configure
make && make clean
cd ..
diff -r checkout.orig checkout

If clean should only clean mms, there are still some files left even for mms ...
Nur in mms-1.0.9-patch60/: config.h.
Nur in mms-1.0.9-patch60/: config.mak.
Nur in mms-1.0.9-patch60/: configure.log.
Nur in mms-1.0.9-patch60/libs/termcap: config.cache.
Nur in mms-1.0.9-patch60/libs/termcap: config.log.
Nur in mms-1.0.9-patch60/libs/termcap: config.status.
Nur in mms-1.0.9-patch60/libs/termcap: libtermcap.a.
Nur in mms-1.0.9-patch60/libs/termcap: Makefile.
Nur in mms-1.0.9-patch60/libs/termcap: termcap.o.
Nur in mms-1.0.9-patch60/libs/termcap: tparam.o.
Nur in mms-1.0.9-patch60/libs/termcap: version.o.
Nur in mms-1.0.9-patch60/libs/tinyxml: tinystr.o.
Nur in mms-1.0.9-patch60/libs/tinyxml: tinyxmlerror.o.
Nur in mms-1.0.9-patch60/libs/tinyxml: tinyxml.o.
Nur in mms-1.0.9-patch60/libs/tinyxml: tinyxmlparser.o.
Nur in mms-1.0.9-patch60/libs/tinyxml: xmltest.
Nur in mms-1.0.9-patch60/libs/tinyxml: xmltest.o.
Nur in mms-1.0.9-patch60/po: cs.mo.
Nur in mms-1.0.9-patch60/po: da.mo.
Nur in mms-1.0.9-patch60/po: de.mo.
Nur in mms-1.0.9-patch60/po: es.mo.
Nur in mms-1.0.9-patch60/po: fi.mo.
Nur in mms-1.0.9-patch60/po: fr.mo.
Nur in mms-1.0.9-patch60/po: pl.mo.
Nur in mms-1.0.9-patch60/po: pt.mo.
Nur in mms-1.0.9-patch60/po: ru.mo.
Nur in mms-1.0.9-patch60/po: sv.mo.
Nur in mms-1.0.9-patch60/: termcap-config.log.
Nur in mms-1.0.9-patch60/termio: getch2.o.
Nur in mms-1.0.9-patch60/: test.c.
Nur in mms-1.0.9-patch60/: version.h.



make clean is meant to mean: remove all object files which a users changes could effect. So the output there is basicly ok. I have removed test.c.

Uatschitchun wrote:
after that
cd checkout
make distclean
cd ..
diff -r checkout.orig checkout

Shouldn't distclean set the source completly back, if I get it right?
Nur in mms-1.0.9-patch60/: config.h.
Nur in mms-1.0.9-patch60/: config.mak.
Nur in mms-1.0.9-patch60/: configure.log.
Nur in mms-1.0.9-patch60/libs/termcap: config.cache.
Nur in mms-1.0.9-patch60/libs/termcap: config.log.
Nur in mms-1.0.9-patch60/libs/termcap: config.status.
Nur in mms-1.0.9-patch60/libs/termcap: Makefile.
Nur in mms-1.0.9-patch60/: termcap-config.log.
Nur in mms-1.0.9-patch60/: test.c.
Nur in mms-1.0.9-patch60/: version.h.



Thanks fixed in patch-62.

User avatar
arj
Site Admin
Posts: 2316
Joined: Thu Dec 01, 2005 8:51 pm
Location: Denmark
Contact:

Re: To be done or not to be done ... that's here the questio

Postby arj » Wed Feb 28, 2007 1:27 am

Uatschitchun wrote: :arrow: Could fullscreen=false be made default .. at least for SDL? 800x600 is not a standard resolution nowadays and let first start be a windowed start is much more safer.

:arrow: Could idle_time=1 be made default?


Fixed in patch-63. Thanks!

User avatar
acmelabs
Overlord
Posts: 2365
Joined: Mon Feb 20, 2006 9:18 pm
Location: Germany
Contact:

Re: To be done or not to be done ... that's here the questio

Postby acmelabs » Wed Feb 28, 2007 6:17 am

hendrik wrote:It happens if you are using remote shares as media drive and even worse if these shares are connected via WLAN. At least this is what I notice on my system, can not speak for acmelabs but I guess he has the same in mind.

Yep, it's NFS and I could double the speed, if I'm connected locally, but nevertheless it's too slow. Evern scrolling is much too slow in such big folders. - Unusable.
You've suggested a bunch of very good ideas. Unfortunately I don't think it's going to make into 1.0.8.2. 'cause it's a major surgery.
@Anders:strace log to come, didn't had the time yesterday, sorry.

Regards,

Andreas

User avatar
acmelabs
Overlord
Posts: 2365
Joined: Mon Feb 20, 2006 9:18 pm
Location: Germany
Contact:

Re: To be done or not to be done ... that's here the questio

Postby acmelabs » Wed Feb 28, 2007 6:24 am

arj wrote:Please let me know what you guys think. I convinced myself that the option 2 was the best, but for that we need a new message (=more translations). So if this is the case then I would like to know ASAP. The 1.0.8.2 is very close and I need to send out request for translations soon to be able to get them all back in time.

OK, so we are talking about "Play Now" functionality.
What if after the "Now Played" songs, the original playlist restores itself magicaly? It's a transparent behaviour and you do no harm to any playlists at all.

Regards,

Andreas

User avatar
Uatschitchun
Overlord
Posts: 3189
Joined: Tue Dec 06, 2005 6:55 pm
Location: Germany
Contact:

Re: To be done or not to be done ... that's here the questio

Postby Uatschitchun » Wed Feb 28, 2007 6:49 am

acmelabs wrote:OK, so we are talking about "Play Now" functionality.

So we even could just leave it disalbled per default :twisted:

Lg
Roman

Ps. Thx again for the prompt fixes!

User avatar
acmelabs
Overlord
Posts: 2365
Joined: Mon Feb 20, 2006 9:18 pm
Location: Germany
Contact:

Re: To be done or not to be done ... that's here the questio

Postby acmelabs » Wed Feb 28, 2007 7:07 am

Uatschitchun wrote:
acmelabs wrote:OK, so we are talking about "Play Now" functionality.

So we even could just leave it disalbled per default :twisted:

Lg
Roman

Ps. Thx again for the prompt fixes!

Err..., if it's a transparent behaviour (restore the original playlist after "Play now") there's no reason to turn it off.

Regards,
Andreas

esprit
master
Posts: 503
Joined: Tue Dec 06, 2005 5:50 pm
Location: France
Contact:

Re: To be done or not to be done ... that's here the questio

Postby esprit » Wed Feb 28, 2007 8:29 am

acmelabs wrote:
hendrik wrote:It happens if you are using remote shares as media drive and even worse if these shares are connected via WLAN. At least this is what I notice on my system, can not speak for acmelabs but I guess he has the same in mind.

Yep, it's NFS and I could double the speed, if I'm connected locally, but nevertheless it's too slow. Evern scrolling is much too slow in such big folders. - Unusable.
You've suggested a bunch of very good ideas. Unfortunately I don't think it's going to make into 1.0.8.2. 'cause it's a major surgery.
@Anders:strace log to come, didn't had the time yesterday, sorry.

Regards,

Andreas

What's your network speed ?

I do NFS access for my Music folders, there are more then 300 folders, quite each one has a cover in it, and it only takes 2 or 3 seconds max the first time I'm going in :shock:

Even konqueror take a lot much time to display the contents of the folder (as mms seems to get the cover only when it is first display, what konqueror don't).

Which NFS version are you using ? are you in UDP or TCP mode ? have you tried to customize your mounting parameters ?

User avatar
acmelabs
Overlord
Posts: 2365
Joined: Mon Feb 20, 2006 9:18 pm
Location: Germany
Contact:

Re: To be done or not to be done ... that's here the questio

Postby acmelabs » Wed Feb 28, 2007 8:40 am

Thanks for your help, I'll open a new topic on that. But I don't want to discuss it here in Romans thread, since he can get very engry, if something is off topic, especially when we're all a bit under preassure due to finalizing 1.0.8.2 and the c't article.
Regards,
Andreas

User avatar
Uatschitchun
Overlord
Posts: 3189
Joined: Tue Dec 06, 2005 6:55 pm
Location: Germany
Contact:

Re: To be done or not to be done ... that's here the questio

Postby Uatschitchun » Wed Feb 28, 2007 4:42 pm

and it only takes 2 or 3 seconds max the first time I'm going in

Since Anders has done the last changes to the lstats, it is working real fast for me, too ... even with wireless!

acmelabs wrote:since he can get very engry

Not really ... even not virtualy ... one on one in direct contact is my kind of getting angry :P

Lg
Roman

P.S. Esprit, has acmelabs ever told you how is music is sorted? Could one tell it sorted? :twisted:

P.P.S That's my way of getting angry ... :roll:

User avatar
acmelabs
Overlord
Posts: 2365
Joined: Mon Feb 20, 2006 9:18 pm
Location: Germany
Contact:

Re: To be done or not to be done ... that's here the questio

Postby acmelabs » Wed Feb 28, 2007 5:17 pm

strace log sent to Anders!

Roman, don't you have an opinion regarding the non-invasive-play-now method? (restoring the old one, if "play now" finished).
I think it's an excellent idea, but nobody seams to give a single dawn about it. Strange.

Regards,
Andreas

esprit
master
Posts: 503
Joined: Tue Dec 06, 2005 5:50 pm
Location: France
Contact:

Re: To be done or not to be done ... that's here the questio

Postby esprit » Wed Feb 28, 2007 5:28 pm

Uatschitchun wrote:
P.S. Esprit, has acmelabs ever told you how is music is sorted? Could one tell it sorted? :twisted:

Hu... please don't tell me it is some kind of All-in-one (*) ? :shock:

If so it's normal, every application will act like this :lol:
And considering using ext3fs + dir_index or reiserfs could perhaps help "a little" (but not as much than mananing your folders the right way :D )

(*) All music files in one dir

esprit
master
Posts: 503
Joined: Tue Dec 06, 2005 5:50 pm
Location: France
Contact:

Re: To be done or not to be done ... that's here the questio

Postby esprit » Wed Feb 28, 2007 5:30 pm

acmelabs wrote:strace log sent to Anders!

Roman, don't you have an opinion regarding the non-invasive-play-now method? (restoring the old one, if "play now" finished).
I think it's an excellent idea, but nobody seams to give a single dawn about it. Strange.

Regards,
Andreas

Personnaly, I don't use the play_now feature, so I can't tell if this is what people who uses it wants or not.


Return to “1.0.9”

Who is online

Users browsing this forum: No registered users and 1 guest