Back to Home / #mythtv / 2008 / 03 / Prev Day | Next Day
#mythtv IRC Logs for 2008-03-22

---Logopened Sat Mar 22 00:00:50 2008
00:03-!-ahbritto [] has joined #mythtv
00:04-!-gnome42 [] has quit [Remote closed the connection]
00:09-!-gnome42 [] has joined #mythtv
00:25-!-code-er [] has joined #mythtv
00:25-!-code-er [] has left #mythtv []
00:41-!-BathoryQuorthon [] has left #mythtv ["Great Hall Awaits a Fallen Brother"]
00:43-!-dasunt [n=dasunt@unaffiliated/dasunt] has joined #mythtv
00:43-!-dasunt [n=dasunt@unaffiliated/dasunt] has left #mythtv []
00:54-!-kambei [n=kambei@unaffiliated/kambei] has joined #mythtv
01:11-!-gnome42 [] has quit [Remote closed the connection]
01:29-!-kambei [n=kambei@unaffiliated/kambei] has left #mythtv ["."]
03:19-!-firekool [] has joined #mythtv
03:23-!-xris [] has quit []
04:01-!-damion [] has joined #mythtv
04:02<damion>im getting an error of unresolvable dependencies when trying to install mythtv on debian
04:11-!-damion [] has quit [Remote closed the connection]
04:18-!-firekool [] has quit [Read error: 104 (Connection reset by peer)]
04:29-!-firekool [] has joined #mythtv
04:39-!-sveinung [] has joined #mythtv
04:47-!-___sveinung [] has joined #mythtv
04:52-!-___sveinung [] has quit [Client Quit]
04:52-!-___sveinung [] has joined #mythtv
04:52-!-___sveinung [] has left #mythtv []
04:56-!-damion [] has joined #mythtv
05:03-!-superm1 [n=superm1@ubuntu/member/superm1] has quit ["Leaving"]
05:07-!-superm1 [n=superm1@ubuntu/member/superm1] has joined #mythtv
05:13-!-sveinung [] has quit [Read error: 110 (Connection timed out)]
05:13-!-kurre2__4 [] has joined #mythtv
05:21-!-kurre2__2 [] has quit [Read error: 113 (No route to host)]
05:24-!-damion [] has left #mythtv ["Leaving"]
05:41-!-rooau1 [] has left #mythtv []
05:41-!-rooau1 [] has joined #mythtv
06:29-!-clever [] has quit [Connection timed out]
06:51-!-|gunni| [] has joined #mythtv
07:39-!-th__ [] has joined #mythtv
07:39<th__>who was it yesterday, who got mythfrontend to compile under Win?
08:06-!-ccooke [] has quit [Read error: 110 (Connection timed out)]
08:52-!-PointyPumper [i=Pintlezz@] has quit [Read error: 104 (Connection reset by peer)]
09:16<janneg>"Specified key was too long; max key length is 1000 bytes"
09:16<janneg>that's a hardcoded value
09:17<janneg>any reasoning why hostname is mostly declared as varchar(255)
09:17<janneg>that looks exzessive long
09:18<janneg>the storage group table has only 64
09:23<Chutt>64 seems fine
09:23<Chutt>hows bout 128?
09:23-!-greend139 [] has joined #mythtv
09:25<janneg>128 would cause trouble for the storage group table
09:26<janneg>it has an unique key over groupname(32), hostname(64) and dirname(255) which is already 1053 bytes
09:27<janneg>I've shortened the dirname to 235
09:28<hads>janneg: No reason here but 255 is a mysql 4.0 varchar limit.
09:36-!-_gunni_ [] has joined #mythtv
09:51<gbee>64 seems more than enough for a hostname
09:52-!-|gunni| [] has quit [Read error: 110 (Connection timed out)]
10:01<janneg>yes, even for
10:02<Chutt>janneg, still going to just binary, or is it now utf8?
10:03<janneg>utf8, otherwise I wouldn't have problems with the key length
10:03<Chutt>right =)
10:04<janneg>I can't see a not totally ugly way to ensure case insensitve matching with binary
10:05<Chutt>i think it's the way to go
10:05<Chutt>aw, crap
10:05<Chutt>i broke the build at work
10:06<janneg>forking the qt mysql driver doesn't look so unattractive anymore but proper charset in the db has some value too
10:07<Chutt>less maintenance
10:07<Chutt>don't want to have to deal with different versions of qt in a forked driver
10:18-!-greend139 [] has quit []
10:19-!-abarbaccia [] has joined #mythtv
10:38<gbee>mythfrontend -p segfaults
10:39<gbee>little busy with some DIY, but I can look at it tonight if no-one else does
10:42-!-cattelan [] has joined #mythtv
10:53-!-catinpan [] has quit ["Terminated with extreme prejudice - dircproxy 1.0.5"]
10:54-!-PointyPumper [i=Pintlezz@] has joined #mythtv
11:14<th__>janneg, hostname should be 255 chars or less as per RFC 1035.
11:14<th__>so to be IETF-compliant it should really be varchar(255)
11:14-!-abarbaccia [] has left #mythtv []
11:16<th__>"To simplify implementations, the total length of a domain name (i.e., label octets and label length octets) is restricted to 255 octets or less."
11:17<janneg>64 fullfills or less
11:20<gbee>64 is enough, there is a difference between what the standard allows and what is actually used in practice, plus we're not really talking about FQDN here, in the context of mythtv they are unique identifiers, no-one has their mythtv boxes communicating over a WAN
11:22<janneg>strange thing it that it still doesn't work with a utf8 database
11:23<gbee>index length still? or something else?
11:24<janneg>I've removed the conversions from MSqlQuery and set the default charset to utf8
11:24<janneg>no still broken symbols
11:26-!-cattelan [] has quit ["Leaving"]
11:26-!-cattelan [] has joined #mythtv
11:26<janneg>the description seems to be correct
11:33<th__>gbee, seems a bit silly not to save a few hundred bytes and risk a future bug if someone has a long hostname..
11:38<janneg>th__: we aren't trying to save space but avoid mysql's max key size
11:41-!-cattelan_away [] has quit ["Terminated with extreme prejudice - dircproxy 1.2.0"]
11:44-!-reynaldo [] has quit [Read error: 110 (Connection timed out)]
11:48<Chutt>should just use a hostname table
11:49-!-gnome42 [] has joined #mythtv
12:15<th__>janneg, its a small table so why is a key even needed
12:16<janneg>to ensure uniqueness
12:17<th__>how is it a problem if the max key size is 1000 though?
12:18<th__>seems to me it should have group id's and host id's instead of names in that table
12:18<janneg>yes, patch welcome
12:30-!-sc00p [] has quit ["Leaving"]
13:08<gbee>I'd tend to agree that numeric ids are better than using the full name, that applies in several places across the mythtv schema, but replacing it is a job that no-one has stepped forward to do
13:11<gbee>but as I said before, although the column is called 'hostname' that's not really it's purpose for most of what mythtv does - it's just an indentifier and as an identifier it should work just as well truncated to 64 chars (hostnames almost universally put unique identifiers at the start, not the end)
13:12<gbee>we just need a solution right now which allows the QT4 branch to be merged back to trunk, it may not be the solution used when 0.22 is released
13:17<janneg>utf8 database does work now. the prvious error was my own error. I used my awk script on a schema I already played with
13:17<janneg>the program title column was ommitted from the conversion
13:21<janneg>I'm still wondering why it failed though. the mysql server is supposed to convert the data
13:39<Chutt>janneg, doh
13:39<Chutt>david just committed a schema change
13:40<janneg>I've seen it, merging it now to the branch and writing an announcement hopefully avoiding further schema changes until the merge
13:40<Chutt>anything else big remaining, now that you've fixed the utf8 issue?
13:41-!-briand [] has quit [Read error: 110 (Connection timed out)]
13:52<janneg>I can't remember, the mythfrontend -p segfault isn't that important, but I look at it after the merges and after converting my productive system
13:57<gbee>-p segfault can wait, just thought I'd mention it since I'm not able to get much work done this weekend (too busy slicing my hands on plumbing)
14:12-!-mykeul [] has joined #mythtv
14:13<janneg>anyone wants to do the plugin updates?
14:20<Anduin>janneg: I can do them later today
14:39-!-xris [] has joined #mythtv
14:43-!-th1 [] has quit [Read error: 113 (No route to host)]
14:45<janneg>strange, how can oldprogram.oldtitle become non unique due to lossless conversion from utf8 in latin1 over utf8 in binary to proper utf8
14:45-!-kormoc [n=kormoc@unaffiliated/kormoc] has joined #mythtv
14:48-!-briand [] has joined #mythtv
15:06-!-randomhtpcguy [] has joined #mythtv
15:13-!-PointyPumper [i=Pintlezz@] has quit [Read error: 104 (Connection reset by peer)]
15:13-!-PointyPumper [i=Pintlezz@] has joined #mythtv
15:14-!-briand [] has quit [Read error: 110 (Connection timed out)]
15:15-!-briand [] has joined #mythtv
15:33-!-Chase [] has joined #mythtv
15:34<Chase>I want to develop an iphone remote application for mythtv, so I'd like to have program guide info
15:35<Chase>but the iphone doesn't provide any mysql client libs
15:35<Chase>is there a reason there's no mythtv protocol for requesting program guide data from the backend?
15:36-!-briand [] has quit [Read error: 110 (Connection timed out)]
15:36<Chase>or if not, is there anything I should consider in extending the protocol to include this information?
15:39<kormoc>you could use the xml interface to get it
15:39<kormoc>or mythproto
15:39<kormoc>not sure where you get the idea of not being other ways to get at it
15:40-!-briand [] has joined #mythtv
15:40<Chase>what xml interface are you alluding to?
15:41<Chase>and I'm talking about mythproto, it doesn't include a way to get the guide data
15:42<kormoc>sure it does
15:42<Chase>what command are you referring to?
15:42<kormoc>Chase, xml info
15:43-!-clever [] has joined #mythtv
15:43<Chase>do you know when this was added?
15:44<janneg>Chase: or use mythfrontend network control interface
15:44<Chase>janneg, I'm still running 0.20 and I didn't see anything there
15:45<Chase>is there some new information available in 0.21?
15:47<kormoc>you can get program info with QUERY_RECORDING TIMESLOT chanid startime
15:47<kormoc>that's per program
15:47<Chase>I want to get all the guide data
15:47<Chase>and channel data
15:47<Chase>without knowing what programs there are in advance
15:50<Chase>I think the xml interface will work
15:50<kormoc>you could just loop over every starttime in the section of time you want, getting program data for each program with that
15:50*kormoc shrugs
16:10-!-Chase [] has left #mythtv []
16:18-!-turbo [] has joined #mythtv
16:20-!-briand [] has quit [Read error: 110 (Connection timed out)]
16:35-!-JoeBorn [] has quit [Read error: 110 (Connection timed out)]
16:37-!-TelnetManta [] has joined #mythtv
16:47-!-Freeaqingme [] has joined #mythtv
16:55-!-kormoc [n=kormoc@unaffiliated/kormoc] has quit []
17:05-!-MediAWak3 [] has joined #mythtv
17:13<janneg>does someone see why a conversion to utf8 from binary could fail for
17:14<janneg>with "Duplicate entry 'Espana en Comunidad' for key 1"
17:14-!-Freeaqingme [] has quit [Remote closed the connection]
17:16<MrGandalf>first and third rows?
17:18<MrGandalf>sounds like a bad conversion on the part of mysql.
17:20<janneg>I don't know, testing it now. it shouldn't do any conversion for the varbinary to varchar utf8 change
17:20<janneg>and the strings are valid utf8 strings
17:23<janneg>yes, 1 and 3. hmm, it might be maybe collate dependent
17:25<gnome42>janneg: That symptom sounds very similar to the silent breakage I saw after the mysql 5 upgrade. Repair/optimize etc. showed no problems with anything but that duplicate key issue seemed very suspicious to me. Just an outside possibility I guess ;)
17:28<janneg>im trying now COLLATE utf8_bin
17:30<janneg>doesn't help
17:32<gnome42>janneg: Do you happen to get the same problem on the original table (before conversion) if you completely drop the indexes and recreate them? (that was what I ended up doing to see which ones got fubarred)
17:36-!-briand [] has joined #mythtv
17:38-!-turbo [] has quit [Read error: 110 (Connection timed out)]
17:45<janneg>it's working with COLLATE utf8_bin in the ALTER TABLE ... MODIFY ... statement
17:46<janneg>not sure if we want utf8_bin collation though
17:55<Chutt>doesn't sort properly with that or something?
17:59<janneg>yes, it is case sensitive
18:01<janneg>it looks like pure madness to use the collation for testing uniqueness for keys
18:01<Chutt>should just be binary difference
18:01-!-ToadP [] has joined #mythtv
18:01<Chutt>but, ah well
18:04<MrGandalf>I'm gonna ask a stupid question.. if the newer qt mysql plugin won't support encoding utf8 within latin1, wouldn't a couple wrapper classes work?
18:06<janneg>no, we can't wrap the mysql driver internas and converting the data afterwards is hard/impossible
18:06<MrGandalf>ok. as I said, it was a stupid question :)
18:08<gbee>oldtitle? that part of Bruce's watch list stuff?
18:09-!-briand [] has quit [Read error: 110 (Connection timed out)]
18:10-!-briand [] has joined #mythtv
18:10<janneg>utf8_bin would work if we choose it only for keys. if we depend on db order on such a column we could use GROUP BY c COLLATE utf8_unicode_ci
18:11<janneg>gbee: it's not only the oldprogram table, person has the same problem
18:13-!-czth__ [n=dbrobins@nat/microsoft/x-09690960a8769dd9] has joined #mythtv
18:13<gbee>janneg: ok, I'm just thinking that maybe we should try pruning out duplicates - "ESPAÃA EN COMUNIDAD" and "Espana en Comunidad" are the same program so removing the copy with the older airdate wouldn't lose any valuable data
18:14<gbee>something like DELETE FROM a, b WHERE a.oldtitle=utf8(b.oldtitle) AND a.airdate < b.airdate;
18:14<gbee>syntax isn't correct, but you get the point
18:16<gbee>same thing could apply to the credits table, it's only an issue with converting existing data and shouldn't cause problems when new data is inserted
18:18<janneg>that is correct in this case. but the utf collation treat german umlauts as equal to the base character and ß as equal to s or ss which can cause real conflicts
18:19<Chutt>that's silly
18:20<Chutt>there isn't a way to make LIKE/etc case insensitive, but keep the collation utf8_bin?
18:20<gbee>this utf8 problem is starting to get annoying ...
18:24<janneg>... oldtitle COLLATE utf8_general_ci LIKE "z%" ORDER BY oldtitle COLLATE utf8_general_ci;
18:24<janneg>does it
18:24<Chutt>ugly :/
18:25<MrGandalf>Ok, I'm going to throw something else out. How about not using the plugin at all instead using mysql directly?
18:26<Chutt>that's basically completely rewriting a large chunk of myth.
18:26<MrGandalf>I appologize of that's another stupid question.
18:27<MrGandalf>chutt: doesn't have to be, but you'd likely recreate large portions of the plugin
18:28<Chutt>how would that be any better than just forking the plugin internally?
18:28<janneg>setting the connection_collation doesn't help
18:29<MrGandalf>chutt: it would be outside qt and therefore not dependant on qt versions.
18:29<Chutt>MrGandalf, please, no more suggestions.
18:29<MrGandalf>ok, my appologies.
18:30-!-czth_ [n=dbrobins@nat/microsoft/x-ced885007256d1f5] has quit [Read error: 110 (Connection timed out)]
18:33-!-firekool [] has quit [Read error: 104 (Connection reset by peer)]
18:39<janneg>it doesn't seem to be possible to set a different collation just for the key
18:57<gbee>janneg: are credits and oldprogram the only affected tables? (the only ones using a unique key on varchar?)
18:59<janneg>yes, oldprogram.oldtitle and are the only columns causing troubles in my db
18:59<janneg>but there are more tables with UNIQUE on varchars
19:11-!-mykeul [] has quit ["Leaving."]
19:11<gbee>ok, I don't have any good ideas right now, so I'll just keep quiet
19:14-!-_gunni_ [] has quit ["KVIrc 3.2.4 Anomalies"]
19:17<gbee>clever: ok, guess I need to find time to work out why activebuffer is NULL
19:17<gbee>although for now I'd probably just commit another null pointer check
19:18<clever>i suspect it was the massive swaping causing it to skew things again
19:18<clever>it took >30 seconds to even start playing
19:18<clever>and >45 before it reacted to the delete key(by closing the whole window)
19:19<clever>had an ssh open on my lap so it was easy to check for a core
19:19<clever>i could print out other variables
19:20<clever>(gdb) print this
19:20<clever>$1 = (const RingBuffer * const) 0x0
19:20<janneg>we could use views which use the case insensitive collation
19:21<gbee>clever: add anything you think might be relevant to the ticket, I'll only forget and I'm not able to work through the problem this weekend
19:21-!-turbo [] has joined #mythtv
19:21<clever>my source has been moved so gdb cant find it
19:22<clever>all ive confirmed so far is that tv_play.cpp hasnt been modified from the revision
19:22<clever>so its not one of my patches(enless other files are at fault)
19:23-!-briand [] has quit [Read error: 110 (Connection timed out)]
19:24<gbee>can't really type properly atm, trapped one of my fingers between two pipes and I've now got a nasty blood filled blister ...
19:25<clever>my finger joints where hurting lastnight from programing a c++ irc bot for 12+ hours
19:36-!-randomhtpcguy [] has left #mythtv []
20:00-!-ahbritto [] has quit [Client Quit]
20:12<rooau1>Is it desirable to remove all cout and cerr from the code? If so what is the best method, using .arg() to QString? Was there some issue with utf8 or similar?
20:43-!-Adin [] has joined #mythtv
20:56-!-grokky [] has joined #mythtv
20:57-!-Adin [] has quit ["Leaving"]
21:33-!-damion [] has joined #mythtv
21:33<damion>when i try to start my frontend it says it cant find the database
21:34<iamlindoro_>read topic
21:34<damion>what topic?
21:34<iamlindoro_>for this channel-- this is the developer channel
21:35-!-damion [] has left #mythtv []
21:44-!-davilla [] has joined #mythtv
21:56-!-croppa [] has quit ["Konversation terminated!"]
22:32-!-VanessaE [n=vanessa@] has joined #mythtv
22:32-!-VanessaE [n=vanessa@] has left #mythtv ["Leaving"]
23:09-!-turbo is now known as briand
23:25-!-rooau1 [] has quit ["Leaving."]
23:25-!-rooaus [] has joined #mythtv
23:35-!-MediAWak3 [] has left #mythtv []
23:55-!-turbo [] has joined #mythtv
23:57-!-briand [] has quit [Read error: 110 (Connection timed out)]
---Logclosed Sun Mar 23 00:00:26 2008