| 01:00 |
| 01:23:00 | | Quit massiveH (Quit: Leaving) |
| 01:36:07 | *** | Saving seen data "./dancer.seen" |
| 02:00 |
| 02:06:51 | | Quit othello7 (Ping timeout: 246 seconds) |
| 02:22:49 | | Quit paulk-bis (Ping timeout: 260 seconds) |
| 02:26:34 | | Join paulk [0] (~paulk@vpn-0-22.aquilenet.fr) |
| 02:26:34 | | Quit paulk (Changing host) |
| 02:26:34 | | Join paulk [0] (~paulk@about/aquilenet/user/paulk) |
| 03:00 |
| 03:30:48 | | Join lebellium [0] (~lebellium@2a01cb0405d07f00d33a0da62bfc2206.ipv6.abo.wanadoo.fr) |
| 03:36:10 | *** | Saving seen data "./dancer.seen" |
| 04:00 |
| 04:34:49 | | Quit PheralSparky (Quit: Leaving) |
| 05:00 |
| 05:10:45 | | Quit jjs0077018310196 (Ping timeout: 248 seconds) |
| 05:19:09 | | Join jjs0077018310196 [0] (~jjs007@host86-191-158-3.range86-191.btcentralplus.com) |
| 05:36:15 | *** | Saving seen data "./dancer.seen" |
| 07:00 |
| 07:36:19 | *** | No seen item changed, no save performed. |
| 08:00 |
| 08:04:14 | | Quit Tonux (Ping timeout: 252 seconds) |
| 08:25:42 | | Join Tonux [0] (~Tonux@193.32.127.222) |
| 08:54:58 | rb-bluebot | Build Server message: New build round started. Revision 1e7334a22c, 345 builds, 9 clients. |
| 08:54:59 | rb-bluebot | manual: Document the 128GB AppleOS limit of early ipod6g models. by Solomon Peachy |
| 09:00 |
| 09:11:03 | rb-bluebot | Build Server message: Build round completed after 966 seconds. |
| 09:11:05 | rb-bluebot | Build Server message: Revision 1e7334a22c result: All green |
| 09:21:02 | rb-bluebot | Build Server message: New build round started. Revision e5d4e33403, 345 builds, 10 clients. |
| 09:21:03 | rb-bluebot | manual: Additional notes about Flash/SSD mods by Solomon Peachy |
| 09:34:40 | rb-bluebot | Build Server message: Build round completed after 820 seconds. |
| 09:34:42 | rb-bluebot | Build Server message: Revision e5d4e33403 result: All green |
| 09:36:21 | *** | Saving seen data "./dancer.seen" |
| 09:40:13 | speachy | I updated the old iFlash topic on the forum to point at the manual, and locked it. |
| 09:45:31 | speachy | When doing that port to SDL2, I had to fix up a couple of things that were specific to MAEMO and PANDORA targets. Anyone know anything about their general state of affairs? |
| 09:51:32 | speachy | given these things are both kinda general purpose Linux, it's not inconveivable to expect SDL2 |
| 09:56:17 | speachy | maemo upstream appears to be effectively dead upstream, can't find anying about SDL2 there. |
| 09:56:44 | speachy | Pandora upstream is effectively dead too, though it has SDL2 at least. |
| 10:00 |
| 10:41:39 | rb-bluebot | Build Server message: New build round started. Revision 8f8fb13edb, 345 builds, 10 clients. |
| 10:41:40 | rb-bluebot | dbtool: Fix build when we have HOTSWAP but not MULTIVOLUME by Solomon Peachy |
| 10:47:53 | speachy | great, the winsim doesnt' work right, not respoding to input events. sigh. |
| 10:53:06 | speachy | so I guess SDL2 still has the same problem about the input events need to being on the same thread that initializes things. |
| 10:55:13 | rb-bluebot | Build Server message: Build round completed after 814 seconds. |
| 10:55:14 | rb-bluebot | Build Server message: Revision 8f8fb13edb result: All green |
| 11:00 |
| 11:36:24 | *** | No seen item changed, no save performed. |
| 12:00 |
| 12:00:28 | | Quit bpye (Quit: Ping timeout (120 seconds)) |
| 12:01:04 | | Join bpye [0] (~bpye@user/bpye) |
| 12:52:55 | | Join othello7 [0] (~Thunderbi@pool-100-36-176-164.washdc.fios.verizon.net) |
| 12:53:23 | _bilgus | I've been battling this make file for multiple hours now and I still can't seem to figure out why its not bringing my plugin defines in, its missing the define for SIMULATOR which causes all kinds of hell to break loose |
| 12:54:54 | _bilgus | I was under the impression that the plugin.make file took care of that for me maybe I'm missing something |
| 13:00 |
| 13:36:25 | *** | Saving seen data "./dancer.seen" |
| 13:45:02 | | Join Moriar [0] (~moriar@107-200-193-159.lightspeed.stlsmo.sbcglobal.net) |
| 14:00 |
| 14:49:04 | | Join PheralSparky [0] (~S|h|a|w|n@user/shawn/x-4432647) |
| 14:56:32 | speachy | _bilgus: working on espeak? or something else? |
| 14:56:44 | speachy | (and yeah, SIMULATOR is definitely passed through |
| 15:00 |
| 15:04:38 | _bilgus | espeak, I got that figured out only to find out I still have lots of work to do to get rid of malloc even in this older version |
| 15:05:58 | _bilgus | at this point I might as well start at their latest since when i'm done its going to be totally custom to us and hard to update again |
| 15:06:38 | _bilgus | why bother defining extensible interfaces if you are just going to bake in assumptions anyway |
| 15:08:39 | _bilgus | luckily we have TLSF to play malloc |
| 15:10:31 | _bilgus | oddly re the makefile I literally deleted it rewrote it a second time and it worked so I don't know if there was an ordering issue or some stray character |
| 15:11:02 | speachy | I once had a semicolon hiding on column 81 |
| 15:11:27 | speachy | of a for loop declaration. "wtf is this only running once?" |
| 15:11:37 | speachy | and a non-wrapping editor. :( |
| 15:11:56 | bertrik | makefiles can be tricky in the sense that they use TABs |
| 15:12:05 | bertrik | and you cannot just replace them by spaces |
| 15:12:07 | bertrik | IIRC |
| 15:12:29 | bertrik | it can be hard to spot the difference, depending on the editor |
| 15:12:32 | _bilgus | ah that could be it I bet when I deleted it my editor used tabs where as maybe the other was using spaces |
| 15:12:56 | speachy | would it make more sense to say "alphabetically" instead of "by first letter" ? |
| 15:13:27 | speachy | fnially working on that db i18n/voicing stuff |
| 15:13:44 | _bilgus | yeah thatd make sense I probably started the file by hiting space and then it decided to use spaces over tabs |
| 15:16:05 | _bilgus | Idk on the strings Alphabetically might imply that abba comes before ac/dc where as first letter makes no such claim? pretty sure the latter is our behavior though |
| 15:16:28 | _bilgus | sorry FORMER** |
| 15:18:23 | speachy | I don't know if we case fold properly even for English |
| 15:18:32 | speachy | we certianly don't for other languages |
| 15:19:11 | speachy | it used to be "A-Z", OlsoFR changed it to be "First Letter" which is rather awkward in some of the other combinations. |
| 15:19:39 | _bilgus | A-Z is going to be harder in other languages |
| 15:20:07 | speachy | that's why I think alphabetic is a better start |
| 15:20:22 | speachy | since that is the intent (if not necessarily implementable in other languages currently) |
| 15:20:28 | _bilgus | sounds good to me in english at least it holds true |
| 15:21:12 | speachy | (if we make the db dynamically generate the "set of first letters in the category" list it'll work for everyone. |
| 15:21:30 | speachy | (case folding notwithstanding..) |
| 15:23:45 | speachy | "Artist A to Z" became "Artists by First Letter" −− so that might become "Artists, Alphabeticaly" ? |
| 15:25:11 | _bilgus | too bad we don't have a way to reuse string in others easily |
| 15:26:09 | speachy | well we _could_ but in this context it's going to be a challenge due to different languages conjucating phrases differently. plus the tree system wouldn't let us voice any combined phrases like that |
| 15:26:19 | _bilgus | suppose that would just make multilingual even more cumbersome |
| 15:27:30 | _bilgus | bah, anyway I'm gonna stash this espeak thing and take a look at the talkclip stuff |
| 15:27:31 | speachy | given multilingual is the entire point... |
| 15:28:08 | speachy | I'm working on a squashed set from those two patch sets. |
| 15:28:46 | speachy | only a minor change from what you posted last (re-ordering the talk_id lookup to come first) |
| 15:29:10 | speachy | but I want to get these language strings/phrases working a bit better for English before I merge. |
| 15:30:59 | speachy | I temporarily disabled the talk_clip lookup too since that's going to take a bit of work to sort out. I want to have a tool to extract the phrases out of the database, so it can all be done in a single commit. |
| 15:31:49 | _bilgus | no worries I just want to have a look at how the filenames come through the voice interface feel free to push your changes back to the same branch and i'll pull them before I push anything new |
| 15:31:59 | speachy | meaning I want the feature to be complete and usable when it lands; as it stands we'll pay the price of looking up a nonexistant talkclip for every line in the db browser. |
| 15:33:59 | speachy | then there's stuff like LANG_PROPERTIES_FILENAME vs LANG_RECORDING_FILENAME and a new LANG_FILENAME in this patch set |
| 15:34:06 | _bilgus | but only when selected* |
| 15:34:07 | speachy | differing only in punctuation |
| 15:35:10 | _bilgus | figure you only look up the talk clip when you select the item I don't know that it'll be better doing that upfront |
| 15:36:10 | speachy | yeah, only when selected, but consider the common case of scrolling down in a list. :) |
| 15:36:29 | *** | No seen item changed, no save performed. |
| 15:38:16 | _bilgus | yeah if slow enough I suppose pretty sure once it speeds up its one two skip a few |
| 15:38:53 | _bilgus | in that case you should probably add a scan for talk clips option |
| 15:39:28 | _bilgus | maybe it wouldn't be so bad to just scan the db folders at init |
| 15:40:07 | speachy | add a flag to the entry to say whether or not there's a clip? |
| 15:40:31 | _bilgus | thats scanning every entry at init |
| 15:41:00 | speachy | yeah |
| 15:41:00 | _bilgus | wonder if we check the db for valid files currently |
| 15:41:04 | speachy | we don't. |
| 15:41:17 | _bilgus | if so itd make sense to in the same op |
| 15:43:17 | _bilgus | what if we just scan the relevant db paths for .talk files and when we find one do a db lookup and save the index |
| 15:43:35 | _bilgus | flip the scan basically |
| 15:43:53 | speachy | welcome to the advantage of the likes of itunes managing all of this for us |
| 15:44:12 | speachy | too bad we can't monetize all of this crap and retire |
| 15:44:29 | _bilgus | it wouldnt be as much fun |
| 15:44:58 | speachy | heh, I bet if we made the next release "Rockbox AI 5G edition" we could get cash thrown at us |
| 15:45:04 | _bilgus | ok so we just do it in the db commit |
| 15:45:24 | _bilgus | if the user has new talk clips they do a recan |
| 15:45:31 | _bilgus | rescan |
| 15:45:43 | speachy | now that's a decent idea |
| 15:45:55 | _bilgus | pretty easy to throw a new file index at it |
| 15:46:20 | speachy | of course it's not just one clip per file, it's a clip per db entry. |
| 15:47:09 | speachy | ie a single file would have what, 6 clips in theory? one per db field. |
| 15:47:45 | _bilgus | you want to voice the metadata as well? |
| 15:48:26 | speachy | yeah, that's the entire point. we already have talkclips for the filenames |
| 15:49:02 | _bilgus | so in that case you should probably generate them as a single talkfile and have several in each |
| 15:49:13 | speachy | that's a lot of duplication |
| 15:49:21 | _bilgus | like index0 is artist 1 is album artist |
| 15:49:52 | speachy | ...for a given album, all of the clips except for the title will be the same. |
| 15:49:56 | _bilgus | so is indexing individual clips |
| 15:50:37 | speachy | I wasn't thinking of indexing the clips themselves, just a "does it exist" to save us the trouble of looking on disk for each one. but that's just an optimization. |
| 15:51:56 | speachy | I mean for a given piece of metatata ("artist:angry white chick") there will be a canonical place to look it up (".rockbox/database/clips/artist/angry white chick.talk") |
| 15:52:00 | _bilgus | well you could save flags to the index in the same vein 0x1 is title 0x2 is album 0x4 is artist |
| 15:52:33 | _bilgus | refrence the index to find out if a clip exists then look it up |
| 15:54:59 | speachy | hmm, this isn't finding/voicing the alphabetic entries (eg "A") |
| 15:57:09 | _bilgus | missing voice clip? |
| 15:57:27 | _bilgus | I get the same with 'B' but everything else voices |
| 15:58:00 | speachy | no.. "tagtree_get_entry_name: 'E' id: -1" |
| 15:58:22 | speachy | B shows up with an id (393) but doesn't voice either |
| 15:59:16 | speachy | interestingly, going into the top-level Files causes a segfault! |
| 15:59:39 | _bilgus | thats an un P2STR |
| 15:59:44 | speachy | I wonder if this is an SDL2 situation |
| 16:00 |
| 16:01:08 | _bilgus | no its a missing P2Str for the segfault I bet |
| 16:01:17 | _bilgus | the voicing issue IDK yet |
| 16:01:17 | speachy | on the main rockbox menu? |
| 16:01:55 | speachy | yeah, git master is broken for the simulator. huh. |
| 16:01:56 | _bilgus | I canged a STR to ID2P in there on files |
| 16:02:13 | speachy | oh, I See what you mean |
| 16:02:25 | _bilgus | tree.c 509 |
| 16:03:03 | _bilgus | but I'm pretty sure that means its not covered in the db menu case either |
| 16:03:29 | _bilgus | gotta be in the scroll engine i imagine |
| 16:03:44 | speachy | yeah, that did it |
| 16:04:18 | speachy | That only seems to affect the root menu |
| 16:04:43 | _bilgus | that stopped it for this instance I think you can still hit it through the db menu |
| 16:05:49 | speachy | the seffault is actually in the unicode code |
| 16:06:07 | speachy | utf8decode |
| 16:07:30 | speachy | _bilgus: https://www.shaftnet.org/~pizza/dump.txt |
| 16:08:17 | | Join _bilgus_ [0] (~bilgus@syn-162-154-213-134.res.spectrum.com) |
| 16:08:43 | speachy | looks like gui_synclist_set_title() doesn't like Ps |
| 16:08:51 | _bilgus_ | sorry got DC little further down is gui_synclist_set_title(list, P2STR((unsigned char*)title), icon); |
| 16:09:33 | | Quit _bilgus (Ping timeout: 276 seconds) |
| 16:10:21 | _bilgus_ | its common to both but the db instance is already translated by that point so str( should do it I think |
| 16:11:24 | speachy | oh... I know what's going on |
| 16:11:32 | speachy | with the not-voiced letters |
| 16:11:54 | speachy | they are VOICE-only entries. |
| 16:12:03 | speachy | so there's no string to compare them to |
| 16:12:31 | _bilgus_ | Ah |
| 16:12:54 | speachy | </grumble> |
| 16:13:06 | _bilgus_ | realtively easy fix |
| 16:13:34 | _bilgus_ | so voice only index should allow you to map A- letter amd get the index |
| 16:13:52 | _bilgus_ | or a whichever case it defualts to |
| 16:16:17 | speachy | first we have to make sure the entry is a single letter. :) |
| 16:16:43 | speachy | g#5938 is the current state of affairs. |
| 16:16:47 | rb-bluebot | Gerrit review #5938 at https://gerrit.rockbox.org/r/c/rockbox/+/5938 : WIP: Database view is fully translatable and entries can be voiced. by Paul Sauro |
| 16:17:37 | speachy | whoops, meant to put my own byline onto that so the message is less confusing. |
| 16:18:39 | _bilgus_ | _talk_spell already has the basic logic |
| 16:19:11 | speachy | trying to figure out _where_ to do this |
| 16:20:24 | _bilgus_ | does that patch supersede the other two then (can abandon it)? |
| 16:22:30 | speachy | yeah, that's the idea. |
| 16:22:38 | _bilgus_ | englist_to_id seems like a good place |
| 16:22:51 | speachy | I'll probably push this over the original one |
| 16:23:02 | speachy | ooh, that's better than what I'd done. let me make sure this works first. |
| 16:26:10 | | Quit Moriar (Quit: Leaving.) |
| 16:26:53 | speachy | oh fml |
| 16:27:15 | speachy | so that fixes the voice, but now it won't display anything because the VOICE_CHAR_* don't have associated strings. :) |
| 16:27:30 | _bilgus_ | lol |
| 16:27:48 | _bilgus_ | yeah because now it uses the ID huh |
| 16:28:00 | speachy | yep |
| 16:29:07 | _bilgus_ | thats annoying and pretty sure if you add strings to the voice only ones it causes issues too |
| 16:29:55 | speachy | I think it's harmless to add strings to the voice only ones, beyond wasting a little bit of space. |
| 16:30:10 | _bilgus_ | ok so jkeep that and then add in a check if id > voice only fallback to the buggered sting? |
| 16:30:57 | _bilgus_ | well buggered string sounds interesting but thats fall back to the untranslated string |
| 16:31:45 | _bilgus_ | much prefer adding a string to them if at all possible its only 100 bytes |
| 16:31:57 | speachy | another wrinkle is that "B" and "Q" are already existant |
| 16:32:43 | _bilgus_ | so I wonder why it works for me except B\ |
| 16:32:54 | _bilgus_ | q works and all the rest of A-Z |
| 16:33:05 | speachy | LANG_SYSFONT_EQUALIZER_BAND_Q and LANG_BYTE |
| 16:33:16 | _bilgus_ | AH we just tell it to not translate |
| 16:33:30 | _bilgus_ | if id > Voice only id = -1 |
| 16:33:51 | _bilgus_ | then it'll just spell it |
| 16:33:54 | _bilgus_ | 'A |
| 16:34:09 | speachy | wtf |
| 16:34:19 | speachy | LANG_EQUALIZER_BAND_Q LANG_SYSFONT_EQUALIZER_BAND_Q |
| 16:35:03 | speachy | completely identical |
| 16:37:04 | _bilgus_ | if (lang_id >= 0 && lang_id < TALK_FINAL_ID_VOICEONLY) |
| 16:38:06 | speachy | where? |
| 16:38:19 | _bilgus_ | in tagcache.c |
| 16:38:26 | _bilgus_ | the two englisg to id calls |
| 16:39:36 | speachy | so what else might english_to_id() barf on when given the voice-only stuff? |
| 16:39:47 | speachy | I mean letters |
| 16:43:25 | _bilgus_ | skin engine is the other user so.. quagmire |
| 16:43:44 | speachy | very few themes use translatable strings |
| 16:44:04 | speachy | and I highly doubt anyone will use single-letters since that didn't work anyway. :D |
| 16:44:59 | _bilgus_ | LANG_RECORDING_FILENAME? |
| 16:45:07 | rb-bluebot | Build Server message: New build round started. Revision 77ff799502, 345 builds, 9 clients. |
| 16:45:07 | rb-bluebot | lang: Delete two completely redundant strings by Solomon Peachy |
| 16:45:14 | _bilgus_ | ah |
| 16:45:41 | speachy | yeah I got rid of RECORDING_FILENAME, fixed it up to use the enw LANG_FILENAME instead |
| 16:48:41 | speachy | ok, pushed it over the original, incorporating everything so far. |
| 16:54:21 | _bilgus_ | still getting an error about it maybe ccache |
| 16:54:33 | speachy | or out-of-date dependencies |
| 16:54:45 | speachy | (make dep!) |
| 16:54:53 | _bilgus_ | tried that already |
| 16:55:26 | speachy | what target? |
| 16:55:34 | _bilgus_ | clipzipsim |
| 16:55:45 | speachy | wait, I missed some. |
| 16:55:51 | speachy | my target didn't have recording. |
| 16:56:11 | _bilgus_ | ah |
| 16:56:36 | _bilgus_ | wish I had thought of that before I deleted the cache :P |
| 16:56:39 | speachy | er, no I didn't. I must have accidently dropped that from my merge. |
| 16:58:51 | speachy | ok, pushed an updated patch, should be good now. |
| 17:00 |
| 17:01:17 | rb-bluebot | Build Server message: Build round completed after 971 seconds. |
| 17:01:19 | rb-bluebot | Build Server message: Revision 77ff799502 result: All green |
| 17:04:42 | speachy | this database i18n/voicing patch represents such a huge step forward, the database is among the few i18n/voice fails we still have. |
| 17:05:22 | speachy | (the other notable fail is spelling out non-english stuff) |
| 17:05:58 | _bilgus_ | thats still better than nothing though |
| 17:06:52 | _bilgus_ | ah I have to do a new voice too |
| 17:07:32 | speachy | with the on-by-default voice pools, make voice is awesome. :D |
| 17:07:56 | _bilgus_ | was just thinking that |
| 17:08:27 | speachy | even with burns-all-the-cores piper doing the voicing. |
| 17:13:04 | _bilgus_ | there were some items the db didn't have that pushed Chris_s to read metadata from disk, if we are expanding the db index maybe we should sve that info as well |
| 17:13:33 | _bilgus_ | frequency was one of them |
| 17:14:50 | _bilgus_ | file size, codec |
| 17:18:19 | | Quit wLLm (Remote host closed the connection) |
| 17:19:05 | speachy | does that really matter in most contexts? I mean it's not like you're going to browse by file size and codec? |
| 17:19:15 | speachy | (or freq?) |
| 17:19:29 | | Quit lebellium (Quit: Leaving) |
| 17:19:33 | _bilgus_ | thats true well maybe frequency |
| 17:20:50 | _bilgus_ | well actually there is probably someone who wants to sort on each of those at some point but yeah no point in overhead for it |
| 17:21:45 | _bilgus_ | same is true of talk clipss though so maybe they should be db adjacent |
| 17:22:19 | _bilgus_ | talk_clip idx and use the same format but don't tie it into the actual db |
| 17:23:51 | _bilgus_ | like the virtual tags |
| 17:26:00 | _bilgus_ | now I have no letters |
| 17:29:22 | speachy | wha I committed worked for me fwiw |
| 17:36:26 | _bilgus_ | did you try enabling logf? it segfaults for me now |
| 17:36:33 | *** | Saving seen data "./dancer.seen" |
| 17:42:19 | _bilgus_ | actually as it stands if I browse by first letter then select the folder it seg faults must be missing something |
| 17:46:07 | speachy | damnit now it's now showing up again. |
| 17:48:28 | speachy | that < TALK_FINAL_ID_VOICEONLY is wrong, should be TALK_FINAL_ID |
| 17:48:41 | | Join Moriar [0] (~moriar@107-200-193-159.lightspeed.stlsmo.sbcglobal.net) |
| 17:52:45 | speachy | that solves teh crash but effectively negates hte voice ID being returned. |
| 17:52:54 | _bilgus_ | WITLF |
| 17:53:20 | _bilgus_ | it worked before with >0 and nothing after |
| 17:55:11 | speachy | ...what was I running then? |
| 17:55:15 | speachy | could have sworn I had both. |
| 17:55:44 | _bilgus_ | well shit talk_id_to_engish is dangerous in the skin engine |
| 17:56:08 | speachy | how so? I mean it's been there for some time |
| 17:56:43 | _bilgus_ | if you hit it when will you know that it translated A to a voice only entry |
| 17:57:19 | _bilgus_ | unless its only using it in a talk kinda context everyone needs to check explicitly |
| 17:58:01 | speachy | it's using it purely for display purposes IIRC. |
| 17:58:19 | speachy | I don't tink anything in the skin engine voices anything |
| 17:59:16 | speachy | does P2STR(VOICE_*) return a NULL or \0 ? |
| 17:59:22 | _bilgus_ | in that case it should probably just treat everything VOICE_FINAL_ID and above as -1 |
| 17:59:39 | | Join wLLm [0] (~wLLm@45.83.5.161) |
| 18:00 |
| 18:00:00 | speachy | talk_id_to_english can't ever match anything in vOICE_* because they are purely voiced strings; the 'dest' is blank. |
| 18:00:13 | _bilgus_ | maybe it is is LANG_LAST_INDEX_IN_ARRAY lower than Final id? |
| 18:00:57 | | Join OlsroFR [0] (~OlsroFR@user/OlsroFR) |
| 18:01:02 | _bilgus_ | yep TALK_FINAL_ID LANG_LAST_INDEX_IN_ARRAY and my dumb self removed that check |
| 18:01:21 | _bilgus_ | notte OLSRoFRs patch had it |
| 18:02:26 | OlsroFR | glad to see that this patch is getting closer on getting merged. I will love to see my database fully translated |
| 18:03:04 | OlsroFR | _bilgus_ This was a complex developement in this regard; it's very easy here to create regressions and hit niche cases... |
| 18:04:32 | speachy | so it does seem that the simplest path forward is to just add the 'dest' letters to the VOICE_CHAR_* entries. |
| 18:04:51 | | Quit OlsroFR (Client Quit) |
| 18:05:20 | _bilgus_ | how is that simpler than not translating them? |
| 18:05:40 | speachy | to get them voiced? |
| 18:06:14 | _bilgus_ | they still get voiced by talk_spell |
| 18:06:27 | speachy | only if you have spelling turned on (it's not by default) |
| 18:06:37 | _bilgus_ | oh |
| 18:07:03 | speachy | and having _some_ entries in the db menus voiced but not the rest... inconsistent to say the least. |
| 18:07:26 | _bilgus_ | well then yes but I guess you need to leave them in the voice only part |
| 18:08:09 | speachy | ok, let me try and see what happens. |
| 18:09:39 | _bilgus_ | I bet you are going to need to duplicate if that doesn't work |
| 18:11:31 | speachy | yeah, looks like it. sigh. |
| 18:13:32 | speachy | the tooling strips all non-voice stuff from VOICE_* strings |
| 18:15:48 | speachy | so we'd have to s/VOICE_CHAR/LANG_CHAR/ |
| 18:15:53 | _bilgus_ | Well we should be able to remove the voice only ones in favor of real lang ids |
| 18:15:53 | speachy | or is this going too deep? |
| 18:16:23 | speachy | they're only used an a handful of places fwiw. |
| 18:16:28 | _bilgus_ | we just need to check the logic in talk to make sure we don't break some assumption |
| 18:16:40 | speachy | ok, let me see.. |
| 18:17:18 | _bilgus_ | the alternative is detecting the by first letter and enabling spell |
| 18:17:28 | _bilgus_ | that seems worse |
| 18:18:27 | speachy | yeah |
| 18:18:40 | speachy | and of course all of this only works for english |
| 18:19:37 | speachy | well, that worked. |
| 18:20:17 | speachy | git diff |
| 18:20:36 | _bilgus_ | do numbers have the same issue then? |
| 18:20:49 | speachy | we voice those differently |
| 18:21:20 | speachy | there's a 'numeric' menu entry in the database than handles those |
| 18:22:40 | speachy | if this is worth doing then I want to commit it separately. |
| 18:26:29 | _bilgus_ | hope you put it before LANG_BYTE |
| 18:26:40 | _bilgus_ | thats why B didn't work for me |
| 18:27:43 | speachy | I intend to leave that special case in english_to_id() |
| 18:28:26 | _bilgus_ | Oh how I hate when order is imposed by file position I guess they need to be prior to any other single letter ones too |
| 18:28:37 | | Quit Malinux (Ping timeout: 248 seconds) |
| 18:29:51 | _bilgus_ | yikes leaning towards lets do this dedicated in a menu entry made for by first letter |
| 18:30:14 | speachy | what do you mean? |
| 18:31:37 | _bilgus_ | so tagtree is basically coopting the menu entries to build them on the fly if we break the 'By first letter' out into a separate menu we can enable spelling as we enter |
| 18:32:13 | speachy | that would also enable it being dynamic |
| 18:32:16 | _bilgus_ | or how ever we decide by giving it a talk cb |
| 18:32:31 | speachy | ok.. so.. |
| 18:33:56 | speachy | back out thie VOICE_CHAR_* -> LANG_CHAR_* crap, don't treat "first letter" anything special so only "B" and "Q" get voiced, and instead rewrite how the "first letter" stuff works instead? |
| 18:34:14 | speachy | ...instead |
| 18:34:19 | _bilgus_ | I think thats probably our cleanest option |
| 18:34:19 | speachy | (and instead) |
| 18:34:45 | speachy | if we rewrite this, I think it makes sense to make it properly dynamic, so the letters shown are the ones that exist in the database. |
| 18:35:19 | speachy | and we can voice callback/spell/whatever as appropriate. |
| 18:35:33 | _bilgus_ | fits into what we already want we can just make it a tagnav tag |
| 18:35:51 | _bilgus_ | %BYFIRSTLETTER prhaps |
| 18:36:10 | speachy | yeah |
| 18:36:23 | speachy | I'm backing this crap out of the patch |
| 18:36:59 | _bilgus_ | lang_id < LANG_LAST_INDEX_IN_ARRAY would be all we need for the moment |
| 18:37:34 | speachy | I still say that's redundant as we can't ever match on an empty string |
| 18:37:53 | speachy | (and lang_english_to_id() already enforces that) |
| 18:38:31 | _bilgus_ | I'm confused then how it crashed without it |
| 18:38:56 | _bilgus_ | or with the wrong one or is that one a negative number |
| 18:39:12 | speachy | because I augmented the lang_english_to_id() to look for single-letter strings and return a VOICE_* entry |
| 18:39:25 | speachy | so I'm reverting that too |
| 18:40:41 | _bilgus_ | haha ok I was just looking at it like WTF |
| 18:42:40 | speachy | ok, new rev pushed |
| 18:43:44 | _bilgus_ | it looks like parse_line has most of what we already desire |
| 18:44:21 | speachy | the only real reason to not commit what's here now is my desire to improve the english strings |
| 18:44:41 | speachy | (and hopefully what you're doing will obviate the need for some of them) |
| 18:44:53 | _bilgus_ | now the question is generating the letters from the db |
| 18:45:51 | speachy | could get expensive to generate that on demand. And there's potentially an entire unicode swath of first-letter options. |
| 18:46:28 | _bilgus_ | I think I can get you a dynamic menu in the tagnav in the next day or two |
| 18:46:28 | speachy | maybe those lists can be pre-generated as part of the general database indexes? |
| 18:47:07 | speachy | ok. I'll work on better English for this in the mean time, but I don't want to have other folks start translating things until we're done |
| 18:47:35 | speachy | (I don't think a dynamic menu would make a difference anyway) |
| 18:48:46 | speachy | probably won't work on rockbox stuff any more tonight but I guess I need to add a special case into the SDL2 port for windows stupidity. |
| 18:49:13 | _bilgus_ | I have to head out as well but I'll start on it late eve |
| 19:00 |
| 19:36:36 | *** | Saving seen data "./dancer.seen" |
| 19:39:08 | | Join massiveH [0] (~massiveH@2600:4040:a982:dc00:fc4c:69ea:549c:6a72) |
| 21:00 |
| 21:36:39 | *** | No seen item changed, no save performed. |
| 23:00 |
| 23:01:18 | | Quit skipwich (Remote host closed the connection) |
| 23:04:49 | | Quit massiveH (Quit: Leaving) |
| 23:15:02 | | Quit Moriar (Ping timeout: 265 seconds) |
| 23:36:41 | *** | Saving seen data "./dancer.seen" |
icon identifies that the person is a core developer (has commit access).