[Helix-client-dev] AMD64 porting status
Gregory Wright gwright at real.comOn Dec 23, 2006, at 2:20 PM, Massimo Perga wrote: > Hello All. > > Greg, > - could you please suggest me a server you know is working > perfectly under Linux for x86 ? (I can use VLC, gstreamer and so on). I would suggest using a Heilx server or Real server. I would work on verifying local playback first however. > - About the MP3 file, it works fine using the XMMS version for > AMD64-native. Hmm. OK, I would file a bug and attach the MP3 file if you have rights to it. > - Regarding the patch, have I sign off a particular license ? I am not sure what licenses you might have signed. However, you just need to follow the patch submission guidelines to submit a patch. Should just be a cut and paste. > - Is there a way to know if the tag I checked out is not up-to- > date, or a simple "cvs update" is enough to synchronize to the > latest checked-in version ? a simple 'cvs up' will make sure it is current. 'cvs stat' will show you the branch for the file you are working on, to verify you are on the branch you think you are. > - Have you any query I can run to understand if the issue I'm > experiencing has an already opened bug report ? If not, which are > the most important parameter I should use to create one ? You can just submit it. I am sure we don't have any AMD64 patches in bugzilla. If so, we can take care of them. --greg. > > Warren, All, > in order to best coordinate our tasks you can find me sometimes in > the IRC channel under the alias of neppanepenthes or using my e- > mail address to chat using Google Talk. > > Season's greetings! > > Max (Italy, GMT+1) > > 2006/12/23, Gregory Wright <gwright at real.com>: On Dec 21, 2006, at > 9:55 PM, Massimo Perga wrote: > > > Hello All, > > good news: now I've got a working splay (please check the logs > > attached) even if there're still some issue to be fixed. I finally > > used the target reported below by Daniel and the patch for the > > "va_list" suggested to Warren some days ago. > > > > As I should have some hours in the week-end to be spent on this > > project, could you please let me know: > > 1) which are the next actions you suggest me to do in the porting > > It looks like all the errors left are just missing distribution > libraries. > I would say the first thing to do is to take all of the changes made > so far > and send in a code review to the correct lists. > > Here is the patch submission guild lines: > > https://common.helixcommunity.org/2005/patch/ > > > 2) codecs to be tested and URLs for each one of those > > I don't think we have public servers that contain samples of each > datatype. We > do have a test kit somewhere thought. That kit should contain > samples. You can > also download and install a free server to host your samples on and > test streaming > that way. However, I think first we need to have you start testing > local and > streaming playback with just simple MP3. The player you just built > should be > capable of both. We need to make sure all of that is working before > doing an > exhaustive test of all datatypes. > > In addition to that, we need to get a box set up so that we can > generate the missing > distribution libraries. Without them, you will not have all the > datatypes anyway. > If there are no other build busters then we can just launch a build > on the build > farm (once your fixes are checked in). > > --greg. > > > > > > Thanks, > > Regards, > > Max > > > > 2006/12/21, Daniel Yek <dyek at real.com>: Hi Max, > > > > I'm sorry, I didn't say it clearly enough. When building using > > Ribosome > > with different settings, always start with a fresh directory. > Ribosome > > conditionally checks out parts of source trees in the repositories. > > When > > the directories are already present, I don't think Ribosome would > > correct > > the source code that has been checked out. > > > > You can spot the problem in the log: > > ... > > datatype/rm exists > > datatype/smil/renderer exists > > protocol/transport/rtp exists > > audio/device exists > > datatype/lsd exists > > > > > > The settings: > > export SYSTEM_ID=linux-2.6-glibc23-amd64 > > [0] Set BIF branch (hxclient_1_5_0_cayenne) > > [1] Set Target(s) (splay) > > [2] Set Profile (helix-client-all-defines-nodist) > > > > > > > > -- > > Daniel Yek > > > > > > At 11:22 PM 12/20/2006, Massimo Perga wrote: > > >Hello All, > > > I tried to perform a build using the "nodist" on an already > > populated > > > folder, but it seems the results didn't changed: 2 hours to get a > > build > > > and several libraries missing. Have I to perform a fresh checkout > > or the > > > former checkout (using all-defines) is fine the same ? > > > > > >At the moment I get the splay but I don't get the clntcore.so > > library. > > > > > >You can find logs attached. > > > > > >I hope in your help. :-) > > > > > >Thanks, > > >Regards, > > > Max > > > > > > > > >2006/12/20, Gregory Wright > > <<mailto:gwright at real.com> gwright at real.com>: > > >On Dec 19, 2006, at 2:14 PM, Daniel Yek wrote: > > > > Max is on IRC... > > > > > > > > I'm just going to quickly send out the relevant build > failure; it > > > > is caused by RDT: > > > > > > > > > > > > make[1]: Entering directory `/home/massimo/cayenne/client/core' > > > > ... > > > > > > > > make[1]: *** No rule to make target `/home/massimo/cayenne/ > > > > protocol_dist/rtsp/dbg/rdtclntlib.a', needed by `dbg/ > clntcore.so'. > > > > Stop. > > > > make[1]: Leaving directory `/home/massimo/cayenne/client/core' > > > > make: *** [copy] Error 2 > > > > Time used: 25.27 seconds > > > > ERROR: UNIXCompile(client/core) ERROR: Make failed. > > > > > > > > --- Build System Error ------------------------------------ > > > > Make failed. > > > > ----------------------------------------------------------- > > > > > >Hmm. Looks like the fastest way would be to switch to a nodist > build. > > >Looks > > >like there is no getting around the need for us to have someone > > with - > > >restricted > > >access build up the dist libs. > > > > > >Daniel, I have a AMD64 machine in an office near Milko where we > store > > >some > > >old equipment. If you have time and want to grab it, see Milko for > > >the key > > >to the office. You will need to install an OS and 64-bit dev tools. > > >After > > >we get it setup (given time to do so) we can each telnet into > the box > > >and > > >work on stuff. I might have time to do it remotely here in Oregon > > if it > > >gets setup. > > > > > >--greg. > > > > > > > > > > > > > > > > > > > -- > > > > Daniel Yek > > > > > > > > > > > > At 12:53 PM 12/19/2006, Massimo Perga wrote: > > > >> Hello all, > > > >> this morning (Italian time) I was able to perform a build > but it > > > >> took almost 2 hours (see log attached) probably because of CVS. > > > >> Having it working properly could help us to progress, because > > > >> unfortunately I can't spend more than 2 hours/day for this > task, > > > >> so I can begin to work and continue the day after. > > > >> > > > >> Please let me know if you have any trick. > > > >> > > > >> Thanks, > > > >> Regards, > > > >> Max > > > >> > > > >> 2006/12/19, Gregory Wright > > > >> <<mailto: gwright at real.com > > ><mailto:gwright at real.com> gwright at real.com>: > > > >> > > > >> On Dec 18, 2006, at 4:55 PM, Daniel Yek wrote: > > > >> > > > >> > It still appears that Both Warren and Massimo are having > > problems > > > >> > checking out source code. > > > >> > > > > >> > More inline... > > > >> > > > > >> >> Have you first tried to build: > > > >> >> > > > >> >> target: splay > > > >> >> branch: hxclient_1_5_0_cayenne > > > >> >> profile: helix-client-all-defines > > > >> > > > > >> > > > > >> > Greg, should they start with a minimal profile? Would helix- > > client- > > > >> > all-defines results in attempts to check out distribution > > binaries > > > >> > that are simply not there and make them wait trying to check > > out > > > >> > source code? > > > >> > > > >> The distribution binary checkout will fail. However, this > > should not > > > >> take > > > >> too long to do so. If it does, and is really causing problems > > then I > > > >> would > > > >> guess we can look at a no-dist build. However, I really like > > to get > > > >> people > > > >> to use the all-defines/splay build because it is so well > > tested and > > > >> we build > > > >> it all the time. It helps to localize problems and solutions. > > > >> > > > >> The AMD64 dist tree could be created and filled with empty > > objects. > > > >> That way > > > >> the checkout would go fast. You would still have build > > failures, but > > > >> that is > > > >> ok. > > > >> > > > >> Maybe someone could try a no-dist build real fast.... > > > >> > > > >> --greg. > > > >> > > > >> > > > >> > > > > >> > > > > >> >> Here is your current settings: > > > >> >> > > > >> >> branch: hxclient_1_5_0_cayenne > > > >> >> platform: linux-2.6-glibc23-amd64 > > > >> >> profile: /home/massimo/amd64_helix/build/umakepf/default > > > >> >> target(s): splay_mp3 > > > >> >> > > > >> >> > > > >> >> Notice that your using a 'default' profile. This is > wrong, you > > > >> >> need to use 'helix-client-all-defines'. Also, please change > > > >> >> your target to 'splay'. Not all DLLs will build for you, but > > > >> >> that only means you will not be able to play those datatypes > > > >> >> that did not build. > > > >> > > > > >> > > > > >> >> Build System (V2.4.4 ) > > > >> >> time: Sat Dec 16 04:58:01 2006 > > > >> >> outfile: build.out > > > >> >> branch: hxclient_1_5_0_cayenne > > > >> >> platform: linux-2.6-glibc23-amd64 > > > >> >> distribution/archive from: linux-2.6-glibc23-amd64 > > > >> >> build Type: debug > > > >> >> build options: debug > > > >> >> profile: /home/massimo/amd64_helix/build/umakepf/default > > > >> > > > > >> > > > > >> > Massimo: You appear to still be using "profile: default". > That > > > >> > won't work. It is very tempting to try out all kind of BIF > > files/ > > > >> > profiles available, but you really want to stick with > whatever > > > >> > recommended and don't deviate from it a bit, at least > until you > > > >> > have successful builds. > > > >> > > > > >> > Thanks. > > > >> > > > > >> > > > > >> > -- > > > >> > Daniel Yek > > > >> > > > > >> > > > > >> > At 12:25 PM 12/16/2006, Daniel Yek wrote: > > > >> >> [Removed player-dev until the client core porting is done.] > > > >> >> > > > >> >> Hi Massimo, > > > >> >> > > > >> >> The build.out that you sent to me appear to indicate that > you > > > >> >> checked out 150Cay on top of an existing source tree, most > > > >> >> probably Atlas source tree. That would explain why you are > > getting > > > >> >> hxmedpltfm.so, which I recall is Atlas only. > > > >> >> > > > >> >> The following is the file list that Massimo sent over > > through IRC. > > > >> >> > > > >> >> neppanepenthes: > > > >> >> auddevlib.a audxfadelib.a debuglib.a > > hxmedplyeng.so > > > >> >> audfixptutil.a clntcorelib.a dtcomcontlib.a > > hxnetwksvc.so > > > >> >> audgainlib.a contclntlib.a dtutillib.a hxsdp.so > > > >> >> audlimiter.a contlib.a fileiolib.a hxxml.so > > > >> >> audmixlib.a core.13796 httplib.a > > logutillib.a > > > >> >> audresamplib.a core.13812 hxmedpltfmlib.a > > mp3codecfltpt.a > > > >> >> audsvclib.a core.14347 hxmedpltfm.so > > mp3fformat.so > > > >> >> > > > >> >> mp3lib.a rdttranlib.a splay wrtrlib.a > > > >> >> mp3payld.a rmcommonlib.a sysclntlib.a xmllib.a > > > >> >> mp3render.so rtptranlib.a syslib.a > > > >> >> netioclntlib.a rtsplib.a systranlib.a > > > >> >> netiolib.a runtlib.a utillib.a > > > >> >> netsvclib.a sdplib.a utlclntlib.a > > > >> >> protutillib.a smplfsys.so vidsvclib.a > > > >> >> > > > >> >> This are the files I've in debug directory > > > >> >> I've to go out > > > >> >> but I'll be back in a couple of hours > > > >> >> > > > >> >> > > > >> >> Could you start Ribosome in a fresh directory and make a > fresh > > > >> build? > > > >> >> Thanks. > > > >> >> > > > >> >> > > > >> >> -- > > > >> >> Daniel Yek > > > >> >> > > > >> >> > > > >> >>> At 11:59 AM 12/16/2006, Massimo Perga wrote: > > > >> >>> Hello, > > > >> >>> you can find attached my build.out file. > > > >> >>> > > > >> >>> Thanks, > > > >> >>> Regards, > > > >> >>> Max > > > >> >>> > > > >> >>> -- > > > >> >>> Massimo Perga > > > >> >> > > > >> >> > > > >> >>> Build System (V2.4.4) > > > >> >>> time: Sat Dec 16 04:58:01 2006 > > > >> >>> outfile: build.out > > > >> >>> branch: hxclient_1_5_0_cayenne > > > >> >>> platform: linux-2.6-glibc23-amd64 > > > >> >>> distribution/archive from: linux-2.6-glibc23-amd64 > > > >> >>> build Type: debug > > > >> >>> build options: debug > > > >> >>> profile: /home/massimo/amd64_helix/build/umakepf/default > > > >> > > > > >> > > > > >> >> At 09:23 AM 12/16/2006, Massimo Perga wrote: > > > >> >>> Hello All > > > >> >>> good news: I was able to successfull build the player > > "splay". > > > >> >>> Currently I'm having a trouble with m_pPlayer, in fact > > what I get > > > >> >>> is: > > > >> >>> > > > >> >>> Simpleplayer is looking for the client core at ./ > > hxmedpltfm.so > > > >> >>> Common DLL path DT_Common=. > > > >> >>> Plugin path DT_Plugins=. > > > >> >>> Codec path DT_Codecs=. > > > >> >>> HX_ASSERT failed: (m_pPlayer)... File exerror.cpp , Line 88 > > > >> >>> > > > >> >>> when I try with: > > > >> >>> > > > >> >>> ./splay > > > <<< http://www.booktv.org/ram/feature/1206/ >http:// > www.booktv.org/ > > ram/feature/1206/ > > > >http:// > > > >> < http://www.booktv.org/ram/feature/1206/ >www.booktv.org/ram/ > > feature/1206/ > > > >> >>> > > > btv120906_4c.ra><< http://www.booktv.org/ram/feature/1206/ > >http:// > > www.booktv.org/ram/feature/1206/ > > > > > > >> >http://www.booktv.org/ram/feature/1206/ > > > >> >>> btv120906_4c.ra > > > >> >>> > > > >> >>> Have you any idea which method should I investigate to > > understand > > > >> >>> where that object should be created ? > > > >> >>> Debugging the issue, I've found the assertion fails when > it's > > > >> >>> invoked the method > > > >> >>> > > > >> >>> ppExContexts[i]->Init(ppPlayers[i], pPreferences, > > > >> pszGUID); > > > >> >>> > > > >> >>> in main.cpp:1985. > > > >> >>> > > > >> >>> Thanks in advance for your support, > > > >> >>> Regards, > > > >> >>> Max > > > >> >>> > > > >> >>> 2006/12/14, Greg Wright < > > > >> >>> > > > >> <mailto: > > > > > gwright at real.com><mailto:gwright at real.com><mailto: > gwright at real.com>gwr > > ight at real.com>: > > > >> >>> warren wrote: > > > >> >>> > On Wednesday 13 December 2006 4:50 pm, Massimo Perga > wrote: > > > >> >>> >> Hi All, > > > >> >>> >> I've followed your help and I added the #define as > > suggested > > > >> >>> by Greg. > > > >> >>> >> > > > >> >>> >> As you can see in the log attached, I also miss some > > files: > > > >> >>> Warren, do you > > > >> >>> >> have those files in your environment ? > > > >> >>> > > > > >> >>> > Hi Max, > > > >> >>> > > > > >> >>> > Yes I have those files. > > > >> >>> > > > > >> >>> > FYI, I have noticed that if I do a fresh checkout and > build > > > >> >>> sometimes the > > > >> >>> > server seems to be really slow to respond. Maybe because > > > >> >>> others are doing > > > >> >>> > checkins? > > > >> >>> > > > > >> >>> > Splay is compiling for me but it seg faults when I try > > to play > > > >> >>> an mp3. > > > >> >>> > > > > >> >>> > Greg: When searching Google, I have noticed posting by > you > > > >> >>> stating that > > > >> >>> > compile problems like ours are due to using GCC 4.x and > > that > > > >> >>> helix does not > > > >> >>> > support 4.x. Is that true? > > > >> >>> > > > >> >>> We have ported HEAD to be gcc 4.x compatible. I think > > *most* of > > > >> >>> 150Cay should compile with gcc 4.x now as well (perhaps > some > > > >> >>> problems in the g2mp4 codecs). > > > >> >>> > > > >> >>> Checking out can take a lot of time if your .buildrc file > > > >> references > > > >> >>> code you do not have access too, like the -restricted > > projects. > > > >> >>> > > > >> >>> --greg. > > > >> >>> > > > >> >>> > > > >> >>> > > > > >> >>> > Warren > > > >> >>> > > > >> >>> > > > >> >>> > > > >> >>> > > > >> >>> -- > > > >> >>> Massimo Perga > > > > > > > > > > > > > > > > > > > > <build.out.bz2 > > > _______________________________________________ > > Helix-client-dev mailing list > > Helix-client-dev at helixcommunity.org > > http://lists.helixcommunity.org/mailman/listinfo/helix-client-dev > > > > > -- > Massimo Perga > ______________________________________ > Computer Engineer > > Via Valentino Carrera 61 > 10146 Torino > Italy > > Tel: +393479806753 > ______________________________________ >