From: Miguel F. <mfr...@gm...> - 2004-10-29 23:25:18
|
so... shall we proceed with our cunning plan? :-) the idea was to have the rc7 released this weekend and final 1.0 in about 15 days. therefore rc7 should be our last release candidate supposedly including all major changes that we want to see in 1.0. right now, i can't think of any major patches that have not been merged, except: 1) cue events: i like the idea, but i don't think i will be able to review and test it before 1.0 2) Reinhard's alphablending patches: again, i don't have time to review them. still, imho, it would be nice to have vdr plugin included in 1.0. perhaps we may do this soon in 1.0.2 or something? the testsuite results are looking good. just 1 segfault and 1 deadlock. guenter reporter another segfault. i guess we may fix those between rc7 and 1.0. any other showstoppers please speak now..... miguel |
From: Daniel Caujolle-B. <seg...@cl...> - 2004-10-30 00:12:54
|
Hi, Le Samedi 30 Octobre 2004 01:25, Miguel Freitas a =E9crit=A0: > so... shall we proceed with our cunning plan? :-) > > the idea was to have the rc7 released this weekend and final 1.0 in > about 15 days. therefore rc7 should be our last release candidate > supposedly including all major changes that we want to see in 1.0. > > right now, i can't think of any major patches that have not been merged, > except: 1) cue events: i like the idea, but i don't think i will be able = to > review and test it before 1.0 > 2) Reinhard's alphablending patches: again, i don't have time to > review them. still, imho, it would be nice to have vdr plugin included > in 1.0. perhaps we may do this soon in 1.0.2 or something? > > the testsuite results are looking good. just 1 segfault and 1 > deadlock. guenter reporter another segfault. i guess we may fix those > between rc7 and 1.0. Just for the record, i'm experiencing few problems with seeking, and time= =20 counter seems stales sometimes (DVD / AVI). Cheers. =2D-=20 73's de Daniel "Der Schreckliche", F1RMB. -=3D- Daniel Caujolle-Bert -=3D- seg...@cl... -= =3D- -=3D- http://naboo.homelinux.org -=3D- |
From: Miguel F. <mfr...@gm...> - 2004-10-30 00:46:43
|
On Sat, 30 Oct 2004 02:10:50 +0200, Daniel Caujolle-Bert <seg...@cl...> wrote: > Just for the record, i'm experiencing few problems with seeking, and time > counter seems stales sometimes (DVD / AVI). oh boy. any specific suggestion on how may i reproduce this? what kind of problems with seeking? do they go away with #nocache? Miguel |
From: Daniel Caujolle-B. <seg...@cl...> - 2004-10-30 01:14:53
|
Hi, Le Samedi 30 Octobre 2004 02:46, Miguel Freitas a =E9crit=A0: > On Sat, 30 Oct 2004 02:10:50 +0200, Daniel Caujolle-Bert > > <seg...@cl...> wrote: > > Just for the record, i'm experiencing few problems with seeking, and > > time counter seems stales sometimes (DVD / AVI). > > oh boy. any specific suggestion on how may i reproduce this? > what kind of problems with seeking? do they go away with #nocache? With CVS from yesterday (Thursday), when i start some DVDs, the time cou= nt=20 is blank. With AVI, the seconds aren't incremented, but not all the time.=20 Also, seeking is unresponsible, with AVI. I didn't find a good exemple to reproduce at 100%, sorry. Cheers. =2D-=20 73's de Daniel "Der Schreckliche", F1RMB. -=3D- Daniel Caujolle-Bert -=3D- seg...@cl... -= =3D- -=3D- http://naboo.homelinux.org -=3D- |
From: Bruce R'. M. <br...@yo...> - 2004-10-30 05:02:27
|
On Friday 29 October 2004 18:11, Daniel Caujolle-Bert wrote: > > oh boy. any specific suggestion on how may i reproduce this? > > what kind of problems with seeking? do they go away with #nocache? > > With CVS from yesterday (Thursday), when i start some DVDs, the time > count is blank. With AVI, the seconds aren't incremented, but not all the > time. Also, seeking is unresponsible, with AVI. > > I didn't find a good exemple to reproduce at 100%, sorry. I don't know if this is related but... When playing American Beauty (Special Edition?)on 6a the time counter moves from 0 to 300 seconds at the end of chapter 1. As soon as chapter 2 starts, the time code (xine_get_pos_length) jumps back to 263 seconds and proceeds up from there. I assume that this has something to do with variable bit rates used in differant parts of the movie. Effectively there are two differant parts of the movie which are labeled with time codes 263-300. If I do a time-based seek to 298 seconds, the actual destination I arrive at depends on what is in the buffer cache. There are a few differant DVDs where this effect occurs. I can look up the names on Monday if anyone is interested. The basic problem is that for seeking purposes you want a linear reletionship between time codes and file offset, but that is not possible with a variable bit rate. Without an index file the next best thing would be a monotonically increasing time code but the current method of projecting instantaneous rates leads to temporal artifacts and local extrema. Everything works fine when you play the movie forwards. The problems arise when you try seeking. I'll try to figure out exactly what is going on when I get a chance. I discovered this during my periodic attempts to play movies backwards. BTW: My current attempt at backwards scanning is to leave the xine-engine as it is and tell the decoder/demuxer to deliver frames in reverse order with an inverted polarity on the pts. Does this approach make sense? -b |
From: Miguel F. <mfr...@gm...> - 2004-10-30 12:41:04
|
On Sat, 30 Oct 2004 03:11:43 +0200, Daniel Caujolle-Bert <seg...@cl...> wrote: > With CVS from yesterday (Thursday), when i start some DVDs, the time count > is blank. ok, i reproduced it here. seeking seems to fix time display but slide bar doesn't move... > With AVI, the seconds aren't incremented, but not all the time. > Also, seeking is unresponsible, with AVI. are you using external subtitles? i noticed seeking problems with external subtitles and i just fixed it. commiting now. Miguel |
From: Miguel F. <mfr...@gm...> - 2004-10-30 13:36:09
|
On Sat, 30 Oct 2004 10:41:01 -0200, Miguel Freitas <mfr...@gm...> wrote: > On Sat, 30 Oct 2004 03:11:43 +0200, Daniel Caujolle-Bert > <seg...@cl...> wrote: > > With CVS from yesterday (Thursday), when i start some DVDs, the time count > > is blank. > > ok, i reproduced it here. > seeking seems to fix time display but slide bar doesn't move... fix committed, please give it a try. miguel |
From: Daniel Caujolle-B. <seg...@cl...> - 2004-10-30 22:23:22
|
Hi Miguel, Le Samedi 30 Octobre 2004 14:41, Miguel Freitas a =E9crit=A0: > On Sat, 30 Oct 2004 03:11:43 +0200, Daniel Caujolle-Bert > > <seg...@cl...> wrote: > > With CVS from yesterday (Thursday), when i start some DVDs, the time > > count is blank. > > ok, i reproduced it here. > seeking seems to fix time display but slide bar doesn't move... > > > With AVI, the seconds aren't incremented, but not all the time. > > Also, seeking is unresponsible, with AVI. > > are you using external subtitles? i noticed seeking problems with > external subtitles and i just fixed it. commiting now. Now, i'm not using extsubs. Unfortunately, i can't test this until Monday= =20 evening, i'm far away civilization, but not so far since i can use my GSM=20 modem, but @ 9600bps... too slow for a cvs update, i've already burned 45=20 minutes to retrieve my emails.... Cheers. =2D-=20 73's de Daniel "Der Schreckliche", F1RMB. -=3D- Daniel Caujolle-Bert -=3D- seg...@cl... -= =3D- -=3D- http://naboo.homelinux.org -=3D- |
From: Siggi L. <si...@us...> - 2004-10-30 10:20:18
|
On Fri, 29 Oct 2004, Miguel Freitas wrote: > so... shall we proceed with our cunning plan? :-) I guess we should. Although it won't quite be christmas in 15 days ;-) > right now, i can't think of any major patches that have not been merged, except: > 1) cue events: i like the idea, but i don't think i will be able to > review and test it before 1.0 > 2) Reinhard's alphablending patches: again, i don't have time to > review them. still, imho, it would be nice to have vdr plugin included > in 1.0. perhaps we may do this soon in 1.0.2 or something? The cue points (as well as OS/2 support) would look nice on our press statement about the 1.0 release, as would the VDR plugin. Technically, however, I don't see a reason why they can't be added affter 1.0. > the testsuite results are looking good. just 1 segfault and 1 > deadlock. guenter reporter another segfault. i guess we may fix those > between rc7 and 1.0. The web site update should be complete before 1.0, and it would be good if we could test it with an RC release, which may as well be "RC8", though... (*duck*) > any other showstoppers please speak now..... I've got a few streams (AVI with XviD/mp3) here that make xine stutter and drop audio at certain positions, just as if the CPU werde hopelessly overloaded. However, top shows 50% idle. Unfortunately, the files are 350MB each, so I could only make them available for individual download someplace... Miguel: Will you be the release manager for 1-rc7? Otherwise, I'd be available. Cheers, Siggi |
From: Miguel F. <mfr...@gm...> - 2004-10-30 12:56:39
|
Hi Siggi, On Sat, 30 Oct 2004 12:20:07 +0200 (CEST), Siggi Langauf <si...@us...> wrote: > The web site update should be complete before 1.0, and it would be good if > we could test it with an RC release, which may as well be "RC8", though... > (*duck*) i don't think web site should delay 1.0 in any way. in fact, there were quite some things missing from web site that would (imho) be showstoppers, like lack of developer's information, test suite and bug reporting. but they are all online now, so from a contents point of view we're done. users accessing the web site don't care whether it is xml or mysql ;-) > > any other showstoppers please speak now..... > > I've got a few streams (AVI with XviD/mp3) here that make xine stutter and > drop audio at certain positions, just as if the CPU werde hopelessly > overloaded. However, top shows 50% idle. Unfortunately, the files are > 350MB each, so I could only make them available for individual download > someplace... i might give it a try... > Miguel: Will you be the release manager for 1-rc7? Otherwise, I'd be > available. Siggi, you know you are our best option for release manager... I can only play the "substitute" (like The Who's ;-) what if i try to address these remaining bugs (like dvd time displaying) and you manage the release process? Miguel |
From: Siggi L. <si...@us...> - 2004-10-30 13:56:57
|
Hi Miguel, On Sat, 30 Oct 2004, Miguel Freitas wrote: [...] > i don't think web site should delay 1.0 in any way. in fact, there [...] > users accessing the web site don't care whether it is xml or mysql ;-) Of course they don't care. However, any change is likely to introduce new bug, or maybe just performance issues. And it would be a shame if they occured right with our big 1.0 release. Well, and currently some of the pages are not even valid HTML. One could argue that people don't care, though, as all browsers seem to render the stuff fine, anyway... [...] > Siggi, you know you are our best option for release manager... I can > only play the "substitute" (like The Who's ;-) Thanks a lot for your faith in my abilities. However, I've almost forgotten how to release ;-) > what if i try to address these remaining bugs (like dvd time > displaying) and you manage the release process? Okay, sounds fine to me. I'm going to make some first test builds today and will do the real release tomorrow. As I'm invited to a Helloween party, things will have to be complete around 15:00 UTC, but that shouldn't be a big issue... Cheers, Siggi |
From: Miguel F. <mfr...@gm...> - 2004-10-30 15:10:21
|
On Sat, 30 Oct 2004 15:56:47 +0200 (CEST), Siggi Langauf <si...@us...> wrote: > > users accessing the web site don't care whether it is xml or mysql ;-) > > Of course they don't care. However, any change is likely to introduce new > bug, or maybe just performance issues. And it would be a shame if they > occured right with our big 1.0 release. agreed. that's one more reason to delay the web site change, not the release ;-) > Okay, sounds fine to me. I'm going to make some first test builds today > and will do the real release tomorrow. As I'm invited to a Helloween > party, things will have to be complete around 15:00 UTC, but that > shouldn't be a big issue... ok, so it really sounds like a halloween release! have fun! :-) Miguel |
From: Philipp M. H. <pm...@ti...> - 2004-10-31 10:51:49
|
Hi! On Fri, Oct 29, 2004 at 09:25:11PM -0200, Miguel Freitas wrote: > the testsuite results are looking good. just 1 segfault and 1 > deadlock. guenter reporter another segfault. i guess we may fix those > between rc7 and 1.0. > > any other showstoppers please speak now..... xine CVS from 2004-10-28, Siemens DVB-C with anaglog module, Debian sid, linux_2.6.10-rc1 on Dual-P4 600. 1. I get audio but no video with the new dvb://. 2. Two segfaults with v4l: (gdb) set args --verbose=255 v4l:/Television/182250 (gdb) run Program received signal SIGSEGV, Segmentation fault. [Switching to Thread -1213792128 (LWP 21934)] 0xb55a3123 in xv_set_property (this_gen=0x81ce588, property=0, value=1) at video_out_xv.c:622 622 if (this->deinterlace_enabled (gdb) where #0 0xb55a3123 in xv_set_property (this_gen=0x81ce588, property=0, value=1) at video_out_xv.c:622 #1 0xb7eadaee in vo_set_property (this_gen=0x81cec30, property=0, value=1) at video_out.c:1445 #2 0xb7ebae1e in xine_set_param (stream=0x8714e60, param=16777216, value=1) at xine_interface.c:449 #3 0xb677b9a2 in open_video_capture_device (this=0x8d4cd48) at input_v4l.c:970 #4 0xb677aa63 in v4l_plugin_video_open (this_gen=0x8d4cd48) at input_v4l.c:1674 #5 0xb7e9e0f8 in __open_internal (stream=0x8714e60, mrl=0x8162c18 "v4l:/Television/182250") at xine.c:673 #6 0xb7e9c3f0 in xine_open (stream=0x8714e60, mrl=0x0) at xine.c:1059 #7 0x08050660 in gui_xine_open_and_play (_mrl=0x8162c18 "v4l:/Television/182250", _sub=0x0, start_pos=0, start_time=0, av_offset=0, spu_offset=0, report_error=1) at actions.c:501 #8 0x08050d27 in gui_play (w=0x0, data=0x0) at actions.c:694 #9 0x08057d96 in gui_execute_action_id (action=ACTID_PLAY) at event.c:475 #10 0x0805a856 in on_start (data=0xbfffd2f8) at event.c:1616 #11 0x080cb6a6 in xitk_run (cb=0x805a7de <on_start>, data=0x0) at xitk.c:1957 #12 0x0805acc2 in gui_run (session_opts=0x0) at event.c:1766 #13 0x0806c17a in main (argc=0, argv=0x0) at main.c:2131 (gdb) list video_out_xv.c:622 617 static void xv_compute_output_size (xv_driver_t *this) { 618 619 _x_vo_scale_compute_output_size( &this->sc ); 620 621 /* onefield_xv divide by 2 the number of lines */ 622 if (this->deinterlace_enabled 623 && (this->deinterlace_method == DEINTERLACE_ONEFIELDXV) 624 && (this->cur_frame->format == XINE_IMGFMT_YV12)) { ^^^^^^^^^^^^^^^ is NULL 625 this->sc.displayed_height = this->sc.displayed_height / 2 - 1; 626 this->sc.displayed_yoffset = this->sc.displayed_yoffset / 2; Adding an additional "this->cur_frame &&" solves this, but segfaults later: Program received signal SIGSEGV, Segmentation fault. [Switching to Thread -1387058256 (LWP 23821)] cache_plugin_read_block (this_gen=0x8d8ec30, fifo=0x0, todo=0) at input_cache.c:198 198 this->cur_pos += buf->size; (gdb) list input_cache.c:198 193 } 194 } else { 195 buf = this->main_input_plugin->read_block(this->main_input_plugin, fifo, todo); 196 this->read_call++; 197 this->main_read_call++; 198 this->cur_pos += buf->size; ^^^ is NULL 199 } 200 return buf; 201 } 202 The if-test above checks for NULL in the other case, but not in the else-case. Addingt an "if (buf)" solves this. The resulting video looks strange (not deinterlaced?), no picture available at the moment. BYtE Philipp -- / / (_)__ __ ____ __ Philipp Hahn / /__/ / _ \/ // /\ \/ / /____/_/_//_/\_,_/ /_/\_\ pm...@ti... |
From: Miguel F. <mfr...@gm...> - 2004-10-31 22:22:19
|
On Sun, 31 Oct 2004 11:51:23 +0100, Philipp Matthias Hahn <pm...@ti...> wrote: > 1. I get audio but no video with the new dvb://. Mike Lampard is the dvb maintainer now ;-) > 2. Two segfaults with v4l: > > #2 0xb7ebae1e in xine_set_param (stream=0x8714e60, param=16777216, value=1) at xine_interface.c:449 > #3 0xb677b9a2 in open_video_capture_device (this=0x8d4cd48) at input_v4l.c:970 using XINE_PARAM_VO_DEINTERLACE for deinterlacing is deprecated (although no changes have been made to this particular code, so i wonder how it worked before). we should remove this from v4l. > cache_plugin_read_block (this_gen=0x8d8ec30, fifo=0x0, todo=0) at input_cache.c:198 > 198 this->cur_pos += buf->size; > (gdb) list input_cache.c:198 > 193 } > 194 } else { > 195 buf = this->main_input_plugin->read_block(this->main_input_plugin, fifo, todo); > 196 this->read_call++; > 197 this->main_read_call++; > 198 this->cur_pos += buf->size; > ^^^ is NULL > The if-test above checks for NULL in the other case, but not in the > else-case. Addingt an "if (buf)" solves this. thanks, i will do that. Miguel |
From: Siggi L. <si...@us...> - 2004-10-31 11:33:02
|
On Sun, 31 Oct 2004, Philipp Matthias Hahn wrote: > On Fri, Oct 29, 2004 at 09:25:11PM -0200, Miguel Freitas wrote: > > the testsuite results are looking good. just 1 segfault and 1 > > deadlock. guenter reporter another segfault. i guess we may fix those > > between rc7 and 1.0. > > > > any other showstoppers please speak now..... > > xine CVS from 2004-10-28, Siemens DVB-C with anaglog module, Debian sid, > linux_2.6.10-rc1 on Dual-P4 600. > > 1. I get audio but no video with the new dvb://. Oh dear! Is there anybody with DVB hardware who can have a look at this issue? > 2. Two segfaults with v4l: [...] > The if-test above checks for NULL in the other case, but not in the > else-case. Addingt an "if (buf)" solves this. This looks odd. Unfortunately, adding if(buf) here seems to be more of a workaround than a fix. > The resulting video looks strange (not deinterlaced?), no picture > available at the moment. Which confirms that something else is wrong... I'm afraid the V4L problem is serious enough to hold the rc7 release off. Unfortunately, I don't really have a clou about the V4L input plugin and the (pseudo) demuxer/decoder plugin chain it uses... Any specialists in the house? Cheers, Siggi |
From: Miguel F. <mfr...@gm...> - 2004-10-31 22:32:36
|
On Sun, 31 Oct 2004 12:32:56 +0100 (CET), Siggi Langauf <si...@us...> wrote: > > 1. I get audio but no video with the new dvb://. > > Oh dear! Is there anybody with DVB hardware who can have a look at this > issue? my suggestion: don't hold release because of dvb, rc7 should help us get more testers on the new code... > > 2. Two segfaults with v4l: > [...] > > The if-test above checks for NULL in the other case, but not in the > > else-case. Addingt an "if (buf)" solves this. > > This looks odd. Unfortunately, adding if(buf) here seems to be more of a > workaround than a fix. no, it is fine. the input may return NULL in case of error or something. > I'm afraid the V4L problem is serious enough to hold the rc7 release off. > Unfortunately, I don't really have a clou about the V4L input plugin and > the (pseudo) demuxer/decoder plugin chain it uses... i guess it should be working now. deinterlacing is frontend's job, v4l plugin must not play with these properties... go ahead! ;-) Miguel |
From: Thibaut M. <thi...@gm...> - 2004-11-03 12:05:57
|
On Sun, 31 Oct 2004 20:32:32 -0200, Miguel Freitas <mfr...@gm...> wrote: [...] > i guess it should be working now. deinterlacing is frontend's job, v4l > plugin must not play with these properties... > > go ahead! ;-) nothing new about the release ? > Miguel Thibaut |
From: Siggi L. <si...@us...> - 2004-11-03 23:26:19
|
On Wed, 3 Nov 2004, Thibaut Mattern wrote: > On Sun, 31 Oct 2004 20:32:32 -0200, Miguel Freitas <mfr...@gm...> wrote: > [...] > > i guess it should be working now. deinterlacing is frontend's job, v4l > > plugin must not play with these properties... > > > > go ahead! ;-) > > nothing new about the release ? Sorry, there have been some urgent things at university. I should be able to complete the release tomorrow... Cheers, Siggi |
From: Thibaut M. <thi...@gm...> - 2004-11-08 11:13:47
|
On Thu, 4 Nov 2004 00:25:58 +0100 (CET), Siggi Langauf <si...@us...> wrote: > On Wed, 3 Nov 2004, Thibaut Mattern wrote: > > > > > On Sun, 31 Oct 2004 20:32:32 -0200, Miguel Freitas <mfr...@gm...> wrote: > > [...] > > > i guess it should be working now. deinterlacing is frontend's job, v4l > > > plugin must not play with these properties... > > > > > > go ahead! ;-) > > > > nothing new about the release ? > > Sorry, there have been some urgent things at university. I should be able > to complete the release tomorrow... hmmm ? Do you need some help ? > Cheers, > Siggi Thibaut |
From: Siggi L. <si...@us...> - 2004-11-08 13:02:19
|
On Mon, 8 Nov 2004, Thibaut Mattern wrote: > hmmm ? > Do you need some help ? I guess I need a new laptop, but I managed to make a release tarball, upload is following... Sorry for the delay, Siggi |
From: Miguel F. <mfr...@gm...> - 2004-11-08 13:42:07
|
On Mon, 8 Nov 2004 14:02:02 +0100 (CET), Siggi Langauf <si...@us...> wrote: > I guess I need a new laptop, but I managed to make a release tarball, > upload is following... that nice ppc laptop has died? :( i was about to proceed with the tarball uploading and stuff, but i guess you don't like the autotools i have installed here... Miguel |
From: Siggi L. <si...@us...> - 2004-11-08 17:20:16
|
On Mon, 8 Nov 2004, Miguel Freitas wrote: > On Mon, 8 Nov 2004 14:02:02 +0100 (CET), Siggi Langauf > <si...@us...> wrote: > > I guess I need a new laptop, but I managed to make a release tarball, > > upload is following... > > that nice ppc laptop has died? :( Not exactly died, but it's randomly doing funny things like segfaults or running sed in an infinite loop. Looks like it has a tendency to overheat; I had to put it into the refrigerator for an hour in order to make it buid the whole xine-lib source tree... :-( > i was about to proceed with the tarball uploading and stuff, but i > guess you don't like the autotools i have installed here... You're guessing correctly, unless you have upgraded ;-) Cheers, Siggi |
From: Manfred T. <Man...@ii...> - 2004-11-08 20:00:18
|
Am Montag, 8. November 2004 14:02 schrieb Siggi Langauf: > I guess I need a new laptop, but I managed to make a release tarball, That's bad, my good old PowerBook was broken in July, I know how bad this could be. I've also removed the old SuSE PPC packages (there is no new SuSE PPC box expect the server version and I can't provide new packages), maybe this could be changed at the packman link on the xine page. At the moment there are IA32 (i568 and i686) Packages for SuSE 8.1 (also work down to 7.1), 8.2, 9.0, 9.1 and (new) 9.2. Maybe ix86_64 will follow (not from me, but we have a volunteer at packman). > upload is following... Thanks Siggi. I've seen no announce, but compile is running, it will take a while for all 10 packages... -- Machs gut | http://www.iivs.de/schwinde/buerger/tremmel/ | http://packman.links2linux.de/ Manfred | http://www.knightsoft-net.de |
From: Siggi L. <si...@us...> - 2004-11-08 22:01:09
|
On Mon, 8 Nov 2004, Manfred Tremmel wrote: > > upload is following... > > Thanks Siggi. I've seen no announce, but compile is running, it will > take a while for all 10 packages... SourceForge mail servers have been a bit slow today, but the announcement is on it's way... Cheers, Siggi |