| 00:17:19 | | Join GreenYoshi [0] (~GreenYosh@user/GreenYoshi) |
| 00:27:36 | *** | Saving seen data "./dancer.seen" |
| 00:35:44 | _bilgus | Well I have the crash fixed and the issue with 'Yes' being the menu title is a limitation of P2STR VIRT_PTR ((unsigned char*)0) is set to NULL so when you pass NULL it returns the first lang item whick happens to be 'YES' |
| 00:36:16 | | Quit dconrad (Remote host closed the connection) |
| 00:37:16 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 00:38:37 | _bilgus | I think instead of fixing this I will jus move "Rockbox' to the first lang item and then it'll show that instead |
| 00:41:37 | | Quit dconrad (Ping timeout: 248 seconds) |
| 00:59:39 | | Join GreenYoshi_ [0] (~GreenYosh@user/GreenYoshi) |
| 01:00 |
| 01:01:42 | | Quit GreenYoshi (Ping timeout: 252 seconds) |
| 01:03:09 | | Join GreenYoshi [0] (~GreenYosh@user/GreenYoshi) |
| 01:05:00 | | Quit GreenYoshi_ (Ping timeout: 252 seconds) |
| 01:06:24 | | Quit GreenYoshi (Remote host closed the connection) |
| 01:06:54 | | Join GreenYoshi [0] (~GreenYosh@user/GreenYoshi) |
| 01:08:53 | _bilgus | I guess the build farm missed the gerrit hook |
| 01:09:52 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 01:09:54 | | Quit GreenYoshi (Remote host closed the connection) |
| 01:10:20 | | Join GreenYoshi [0] (~GreenYosh@user/GreenYoshi) |
| 01:14:10 | | Quit dconrad (Ping timeout: 252 seconds) |
| 01:58:26 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 02:00 |
| 02:02:51 | | Quit dconrad (Ping timeout: 252 seconds) |
| 02:17:06 | | Quit martylake (Ping timeout: 248 seconds) |
| 02:20:29 | | Join martylake [0] (~martylake@82.66.174.242) |
| 02:27:39 | *** | Saving seen data "./dancer.seen" |
| 02:37:32 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 02:41:48 | | Quit dconrad (Ping timeout: 252 seconds) |
| 02:45:26 | | Join JanC_ [0] (~janc@user/janc) |
| 02:45:26 | | Nick JanC is now known as Guest7443 (~janc@user/janc) |
| 02:45:26 | | Quit Guest7443 (Killed (zirconium.libera.chat (Nickname regained by services))) |
| 02:45:26 | | Nick JanC_ is now known as JanC (~janc@user/janc) |
| 02:56:05 | | Quit othello7 (Ping timeout: 248 seconds) |
| 03:00 |
| 03:55:46 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 04:00 |
| 04:00:02 | | Quit dconrad (Ping timeout: 248 seconds) |
| 04:01:26 | | Join _bilgus_ [0] (~bilgus@syn-162-154-213-134.res.spectrum.com) |
| 04:02:43 | | Quit _bilgus (Ping timeout: 268 seconds) |
| 04:04:19 | | Quit jacobk (Ping timeout: 260 seconds) |
| 04:05:00 | | Join jacobk [0] (~quassel@47-186-65-73.dlls.tx.frontiernet.net) |
| 04:27:40 | *** | Saving seen data "./dancer.seen" |
| 04:31:00 | | Quit paulk-bis (Quit: WeeChat 3.0) |
| 04:31:14 | | Join paulk [0] (~paulk@about/aquilenet/user/paulk) |
| 04:44:26 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 04:48:31 | | Quit dconrad (Ping timeout: 244 seconds) |
| 04:51:17 | | Join JanC_ [0] (~janc@user/janc) |
| 04:51:17 | | Nick JanC is now known as Guest4906 (~janc@user/janc) |
| 04:51:17 | | Quit Guest4906 (Killed (tantalum.libera.chat (Nickname regained by services))) |
| 04:51:17 | | Nick JanC_ is now known as JanC (~janc@user/janc) |
| 05:00 |
| 05:13:01 | | Quit paulk (Ping timeout: 268 seconds) |
| 05:13:19 | | Join paulk [0] (~paulk@vpn-0-22.aquilenet.fr) |
| 05:13:20 | | Quit paulk (Changing host) |
| 05:13:20 | | Join paulk [0] (~paulk@about/aquilenet/user/paulk) |
| 05:22:01 | | Quit dook (Remote host closed the connection) |
| 05:22:02 | | Quit lane (Remote host closed the connection) |
| 05:22:02 | | Quit ursa-major (Remote host closed the connection) |
| 05:31:48 | | Join JanC_ [0] (~janc@user/janc) |
| 05:31:48 | | Nick JanC is now known as Guest5649 (~janc@user/janc) |
| 05:31:48 | | Quit Guest5649 (Killed (lithium.libera.chat (Nickname regained by services))) |
| 05:31:48 | | Nick JanC_ is now known as JanC (~janc@user/janc) |
| 05:31:51 | | Join OlsroFR [0] (~OlsroFR@user/OlsroFR) |
| 05:32:00 | OlsroFR | https://gerrit.rockbox.org/ the gerrit is throwing me errors 500 |
| 05:32:11 | | Quit OlsroFR (Client Quit) |
| 05:32:56 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 05:33:22 | | Quit Maxdamantus (Ping timeout: 268 seconds) |
| 05:37:48 | | Quit dconrad (Ping timeout: 276 seconds) |
| 05:42:38 | | Join Maxdamantus [0] (~Maxdamant@user/maxdamantus) |
| 05:53:26 | | Quit COMPL_EXE (Read error: Connection reset by peer) |
| 05:53:37 | | Join OriginCode [0] (~compl.exe@aosc/dev/origincode) |
| 05:54:01 | | Nick OriginCode is now known as COMPL_EXE (~compl.exe@aosc/dev/origincode) |
| 06:00 |
| 06:05:21 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 06:05:34 | | Quit q3k (Changing host) |
| 06:05:34 | | Join q3k [0] (q3k@hswaw/infra/q3k) |
| 06:10:24 | | Quit dconrad (Ping timeout: 272 seconds) |
| 06:27:42 | *** | Saving seen data "./dancer.seen" |
| 06:28:26 | | Join OlsroFR [0] (~OlsroFR@user/OlsroFR) |
| 06:28:29 | OlsroFR | it's good now |
| 06:28:31 | | Quit OlsroFR (Client Quit) |
| 06:35:05 | | Join lane [0] (809450f172@2a03:6000:1812:100::1300) |
| 06:35:11 | | Join dook [0] (c9db4a6425@2a03:6000:1812:100::1222) |
| 06:35:12 | | Join ursa-major [0] (114efe6c39@2a03:6000:1812:100::11f3) |
| 06:50:40 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 06:54:54 | | Quit dconrad (Ping timeout: 246 seconds) |
| 06:55:01 | | Quit PheralSparky (Read error: Connection reset by peer) |
| 07:00 |
| 07:08:54 | | Join GreenYoshi_ [0] (~GreenYosh@user/GreenYoshi) |
| 07:11:57 | | Quit GreenYoshi_ (Remote host closed the connection) |
| 07:12:22 | | Join GreenYoshi_ [0] (~GreenYosh@user/GreenYoshi) |
| 07:12:34 | | Quit GreenYoshi (Ping timeout: 248 seconds) |
| 07:16:45 | | Join GreenYoshi__ [0] (~GreenYosh@user/GreenYoshi) |
| 07:20:27 | | Quit GreenYoshi_ (Ping timeout: 252 seconds) |
| 07:25:40 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 07:30:32 | | Quit dconrad (Ping timeout: 268 seconds) |
| 08:00 |
| 08:14:16 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 08:18:45 | | Quit dconrad (Ping timeout: 252 seconds) |
| 08:19:17 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 08:24:10 | | Quit dconrad (Ping timeout: 268 seconds) |
| 08:27:45 | *** | Saving seen data "./dancer.seen" |
| 09:00 |
| 09:02:17 | speachy | _bilgus_: g#3221 |
| 09:02:20 | rb-bluebot | Gerrit review #3221 at https://gerrit.rockbox.org/r/c/rockbox/+/3221 : (Do not merge) FS #9613 - Custom formatting for <All Tracks> DB menus by Christian Soffke |
| 09:02:27 | speachy | whoops |
| 09:02:30 | speachy | g#3321 |
| 09:02:33 | rb-bluebot | Gerrit review #3321 at https://gerrit.rockbox.org/r/c/rockbox/+/3321 : lang: Offset the start of the language VIRT_PTR to avoid a null pointer by Solomon Peachy |
| 09:03:06 | speachy | I wrote it nearly 4 years ago and abandoned it, methinks it shoudl be revived and committed |
| 09:05:03 | speachy | as it turns out, we special-case VIRT_PTR for s5l870x and DM320, but imx31, pp50xx, s3c2440, rk27xx, and tcc780x all have memory at 0x0 |
| 09:05:12 | speachy | s/all/also/ |
| 09:05:42 | speachy | and VIRT_PTR needs to point at an _invalid_ memory location. |
| 09:06:06 | speachy | well, s/memory location/unmapped location in the address space/ |
| 09:07:43 | speachy | aaabbb: AFAICT all of the accidentally-removed-in-3.15 opus optimizations are back in as of some time ago. |
| 09:07:51 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 09:08:32 | speachy | all but one but that was done to free up stack space and didn't make an appreciable difference on the marginal targets. |
| 09:10:21 | speachy | _bilgus_: so we could be looking at some subtle bugs on those targets. |
| 09:12:30 | | Quit dconrad (Ping timeout: 260 seconds) |
| 09:34:47 | speachy | _bilgus_: g#6381 |
| 09:34:50 | rb-bluebot | Gerrit review #6381 at https://gerrit.rockbox.org/r/c/rockbox/+/6381 : internals: VIRT_PTR needs to point at an *invalid* unmapped address by Solomon Peachy |
| 09:35:14 | speachy | revamped that earlier one, and added in all of the "0x0 is real memory" exceptions I could find. |
| 09:35:28 | speachy | it compiles therefore it's good |
| 09:44:26 | speachy | (works on MIPS and simulators so far. I don't have anything else handy) |
| 09:52:41 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 09:56:58 | | Quit dconrad (Ping timeout: 244 seconds) |
| 10:00 |
| 10:14:48 | _bilgus_ | aww I was really enjoying that P2STR(NULL) == 'Rockbox' |
| 10:16:43 | _bilgus_ | I didn' remove them on accident just had no way to test |
| 10:23:04 | speachy | the thing I was worried about is that the compiler might "optimize" away a P2STR(0) reference and cause mayhem. |
| 10:23:25 | speachy | ...since P2STR(0) is legit |
| 10:24:39 | _bilgus_ | well you are for sure right on ID2P though #define ID2P(id) (VIRT_PTR + id) |
| 10:24:50 | _bilgus_ | hat shits the bed |
| 10:27:43 | speachy | the rest of that patch is there to ensure there's no possibility that a "real" pointer would ever point into that 64K reserved revion for the virtual pointers. |
| 10:27:47 | *** | Saving seen data "./dancer.seen" |
| 10:28:05 | speachy | we had special cases for a couple of targets but many more needed something similar |
| 10:28:36 | speachy | During the gcc4.9 uplift I ran into a "0 is real memory" problem with the PP iPods' early init code |
| 10:34:09 | _bilgus_ | It really is an ingenious solution though |
| 10:34:18 | speachy | oh absolutely |
| 10:35:32 | _bilgus_ | we have a few places like that, its always a joy (even when its not lol) |
| 10:37:23 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 10:41:03 | | Quit GreenYoshi__ (Quit: Leaving) |
| 10:41:45 | _bilgus_ | I ran into weird stuff lastnight I was bisecting and a previously merged commit came up but when I tried to build and run it it seg faulted, bisect reset, make clean, make reconf and still seg faulted had to wipe the entire sim disk and I still never figured out what the issue was because when I copied it back it worked fine |
| 10:42:19 | | Quit dconrad (Ping timeout: 268 seconds) |
| 10:44:24 | _bilgus_ | I just chalked it up to something got out of whack but I had two data aborts on start yesterday that cleared by the third boot and no other changes |
| 10:44:37 | _bilgus_ | (at HEAD) |
| 10:45:41 | speachy | so you think my little VIRT_PTR change looks sane? appears to work so far. |
| 10:46:28 | _bilgus_ | I'm building it now.. |
| 10:48:45 | _bilgus_ | adds around 100 bytes and appears to work just fine |
| 10:52:12 | _bilgus_ | I've a feeling that change will probably solve some bugs we didn't know we had |
| 10:53:52 | speachy | that gain is probably from code that was "optimized" out getting put back in |
| 10:54:48 | _bilgus_ | I wonder if anyone has noticed the faster startups yet I do but it might be placebo |
| 10:56:33 | | Join qf [0] (~qf@user/qf) |
| 10:56:34 | speachy | how much faster are we talking about here? |
| 10:56:36 | _bilgus_ | on this device its about a second faster to scrolling the main menu |
| 10:57:32 | _bilgus_ | I noticed it while bisecting |
| 11:00 |
| 11:00:44 | _bilgus_ | most of that is the skin engine that check viewport call was really hot and slow |
| 11:01:34 | _bilgus_ | sorry scan_int and check_vp |
| 11:01:58 | _bilgus_ | scan_int was building a buffer to run atoi on |
| 11:07:59 | _bilgus_ | I guess i'll add faster startup with themes because when I do cabbie if feels snappy either way |
| 11:09:49 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 11:14:10 | | Quit dconrad (Ping timeout: 248 seconds) |
| 11:22:32 | _bilgus_ | and measured its about a half a second faster with cleanscreen spinal tap (argubly pretty hefty) and cabbie is within measurement error of my stop watch clicking speed |
| 11:24:28 | speachy | yesterday I did some more updates on the changelogs |
| 11:26:26 | _bilgus_ | are you putting all the things in there or just the larger aggregate? |
| 11:27:11 | speachy | the majorchanges page is generally noteworthy stuff but the release notes is only the big stuff. |
| 11:29:40 | speachy | certainly open to feedback on the release notes |
| 11:30:54 | _bilgus_ | had to find them https://www.rockbox.org/wiki/ReleaseNotes400.html |
| 11:40:22 | _bilgus_ | looks good little light on the db improvements and should we have something about coverity that was a great many of the bugs |
| 11:43:32 | _bilgus_ | Aww NM its not that much faster the difference I'm seeing in spinal tap is because of the bad background viewport, once I hit a build after g#6315 its no longer so stark |
| 11:43:35 | rb-bluebot | Gerrit review #6315 at https://gerrit.rockbox.org/r/c/rockbox/+/6315 : [Bugfix] Copy current lcd contents into empty background buffers by William Wilgus |
| 11:54:52 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 11:59:07 | | Quit dconrad (Ping timeout: 252 seconds) |
| 12:00 |
| 12:08:37 | speachy | The release notes are more intended to be user-visible stuff, "general bug fixes" is about the only meaningful summary there. |
| 12:15:20 | _bilgus_ | I was saying more like in the thank you part |
| 12:23:59 | speachy | hmm, ought to put that in CREDITS too |
| 12:24:20 | speachy | or no, that's only for code in-tree |
| 12:27:49 | *** | Saving seen data "./dancer.seen" |
| 12:39:23 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 12:44:15 | speachy | ok, another pass at the release notes. |
| 12:44:25 | | Quit dconrad (Ping timeout: 268 seconds) |
| 12:48:29 | speachy | wow, that VIRT_PTR change bumped the binsize by an average of 255 bytes, peak was 416. |
| 12:49:32 | speachy | more than I expected |
| 12:56:19 | _bilgus_ | I'm guessing the wiki aut markup did something with the capital in MacDonald, and deMille, McPhee, BouAntoun, DiFebbo, McIntosh |
| 12:56:41 | speachy | yeah, all of thise CamelCase names need to be prefixed with ! |
| 12:57:56 | speachy | I'd _REALLY_ like to get 4.0 out on, say, March 1st. |
| 12:58:42 | _bilgus_ | why not march 31st into april fools day |
| 12:58:44 | speachy | oh! that reminds me. |
| 12:59:09 | _bilgus_ | drop it at 11:59 3-31 |
| 12:59:32 | _bilgus_ | 4.0 |
| 12:59:40 | * | speachy smacks his forehead. |
| 13:00 |
| 13:03:03 | | Join othello7 [0] (~Thunderbi@pool-100-36-176-164.washdc.fios.verizon.net) |
| 13:06:04 | Xogium | that would be hillarious |
| 13:06:26 | Xogium | people would be like, you're joking, right ? A release of rockbox ? I don't believe you |
| 13:07:40 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 13:12:02 | | Quit dconrad (Ping timeout: 248 seconds) |
| 13:13:30 | _bilgus_ | and yet it would be true :) |
| 13:14:25 | _bilgus_ | I relly need to get entries for somewhere around 10-20 plugins |
| 13:14:40 | speachy | I want to regenerate the full complement of bootloaders |
| 13:14:57 | speachy | to pick up the updated filesystem/disk/etc code |
| 13:15:55 | _bilgus_ | this is going to be or litmus test then |
| 13:16:01 | _bilgus_ | our* |
| 13:16:26 | _bilgus_ | going to be some stuff for 4.01 |
| 13:16:29 | speachy | I know the ipod6g bootloaders work, I can try the mini2g (which will cover the PP ipods). xduoox3 also works. |
| 13:16:59 | _bilgus_ | yeah these are fine along with most of the sansas the rest .. well .. TBD? |
| 13:17:07 | speachy | wish I had some visibility into those recurring "no partition found" errrs folks have when doing the rockbox insteall |
| 13:18:22 | _bilgus_ | Its almost got to be some errant step or bad config I'd think otherwise Olso would have said something about it by now |
| 13:18:23 | speachy | but when said folks manage to consistently ignore instructions... |
| 13:18:36 | speachy | (not to mention consistently leave out important info...) |
| 13:18:43 | _bilgus_ | ^^^^ |
| 13:19:17 | speachy | the ipod5g/6g's funky sector size is a huge source of pain |
| 13:19:32 | speachy | with folks expecting to be able to partition the drives outside the player |
| 13:19:57 | _bilgus_ | I've run into that with the sansas too though TBF |
| 13:20:12 | hactar|ant | 4.0 <3 |
| 13:20:20 | _bilgus_ | you can't let windows do the internal drive or all hell breaks loose |
| 13:20:27 | speachy | not to mention cargo culting "use stock OS for transfers" |
| 13:20:40 | _bilgus_ | thats or own damn fault though |
| 13:20:56 | speachy | and using random patched builds that only succeed accidently |
| 13:21:02 | _bilgus_ | clip+ and zip used to do that too it'll die down with the new release |
| 13:21:26 | speachy | on one hand, that is due to our bugs, on the other hand there's a lot of really fiddly ribbon cables, sd adapters, and sd cards. |
| 13:22:28 | speachy | would it make sense to have rbutil set up the sansas to do multiboot instead? |
| 13:24:15 | _bilgus_ | probably, I figured having the multiboot bootloader installed was probably enough |
| 13:24:49 | speachy | I was thinking that it would be really good to always supply bootdata from our bootloaders, specifying the bootloader revision if nothing else. |
| 13:24:52 | _bilgus_ | doing it from the get go would just step around the problem in the first place |
| 13:25:38 | _bilgus_ | oh so expanding bootdata across the board? |
| 13:25:50 | _bilgus_ | its already in around half |
| 13:27:15 | _bilgus_ | I think what we can do is just make the debug menu always do bootdata and fallback to VER 0 or -1 for unknown stuff |
| 13:27:45 | _bilgus_ | then you know its an old bootloader and start versioning after |
| 13:28:06 | _bilgus_ | Aiden has most of that in already |
| 13:28:56 | _bilgus_ | RN its based on #ifdef MULTI_BOOT and possibly one or two others |
| 13:30:12 | _bilgus_ | basically v0(no bootdata) v1 (no partition data) and v2 (today) |
| 13:30:24 | speachy | I was thinking less abuot an explicit "version" (or API?) number and instead the commit hash |
| 13:30:44 | _bilgus_ | you'd need more than 8 chars |
| 13:30:52 | _bilgus_ | that would make a v3 |
| 13:30:53 | speachy | definitely |
| 13:31:12 | _bilgus_ | be better to have the bootloader pass an address |
| 13:31:31 | speachy | address + API. |
| 13:31:33 | _bilgus_ | to what though .. hmm |
| 13:31:49 | speachy | I mean it's effectively the same thing as bootdata |
| 13:32:01 | _bilgus_ | yeah |
| 13:32:10 | speachy | or the bootdata could just include a pointer to the version string |
| 13:32:24 | _bilgus_ | well yeah but its gone after fw is loaded |
| 13:32:33 | _bilgus_ | thats the whole thing with bd |
| 13:32:46 | _bilgus_ | you and the fw agree to set aside this space |
| 13:32:53 | speachy | true |
| 13:33:09 | speachy | some targets have bootloader in flash, but others are purely RAM-based |
| 13:33:14 | _bilgus_ | we can just extend it by 32 bytes (two rows) and store the string direct |
| 13:34:41 | _bilgus_ | It'll just be V3 anything else that should get or would be nice to get from th bootloader while we are at it? |
| 13:35:28 | speachy | not off the top of my head. this is a nice to have but I don't think it's worth the short-term churn |
| 13:37:17 | _bilgus_ | just having bootdata gives you a cutoff.. |
| 13:37:28 | _bilgus_ | its at least *this* new |
| 13:39:17 | _bilgus_ | I guess I've come to the conclusion that 100,000 tracks is about the limit on the ClipZip |
| 13:39:38 | _bilgus_ | things start feeling really clunky around there |
| 13:39:56 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 13:39:56 | _bilgus_ | mainly just the DB |
| 13:40:56 | _bilgus_ | arguably its been a year of playing this library and I'm still hearing stuff I never have before so its a bit overkill |
| 13:41:36 | speachy | we've scaled remarkably well for 20-year-old hardware and modern TB-sized storage. |
| 13:41:55 | _bilgus_ | it almost makes me want to set it up with that persons patch than puts in station breaks |
| 13:42:28 | _bilgus_ | makes it feel like growing up in the 90's staying up late listening to pirate radio stations |
| 13:44:03 | | Quit dconrad (Ping timeout: 246 seconds) |
| 13:44:58 | _bilgus_ | for a while I was deleting tracks I didn't like but that got old, thats where I think the playback log could come in handy |
| 13:45:15 | speachy | I use mine for little more than audio books |
| 13:45:22 | _bilgus_ | if it was played for less than 1 second delete it |
| 13:45:27 | speachy | though that may change as I try to avoid the news for the next few years |
| 13:45:35 | _bilgus_ | you too? |
| 13:45:48 | speachy | for me it's normally either NPR or audiobooks |
| 13:46:23 | speachy | oh, that reminds me. the next step in a fully voiced database is to extract track metadata into talk clips |
| 13:47:04 | speachy | I'll probably extend the existing dbtool to be able to do that offline |
| 13:47:06 | _bilgus_ | taht will be neat but I almost feel like we should push it into a plugin |
| 13:47:19 | _bilgus_ | and or offline |
| 13:47:31 | speachy | but methinks a plugin to extract that data into a suitably-formatted file is better |
| 13:47:42 | _bilgus_ | figure I have about half of that in the lua TagNav script |
| 13:47:59 | _bilgus_ | it parses the db files for tags |
| 13:48:12 | _bilgus_ | then lets you build custom track classifiers |
| 13:48:42 | speachy | and then the make talkclips cmdline (and/or rbutil) can consume that to generate the appropriate clips |
| 13:48:45 | _bilgus_ | just a bit further to have it generate whole tagnav |
| 13:49:40 | _bilgus_ | its really pretty neat how easy lua makes doing that |
| 13:50:05 | _bilgus_ | its mainly just the regex engine and free-er strings |
| 13:50:06 | speachy | yeah. I figure something simple, like 'type<tab>utf data' |
| 13:50:37 | speachy | and the genered clips coudl live under .rockbox/dbtalk/type/blablabla.talk |
| 13:50:53 | _bilgus_ | I actually have considered pulling out the string parsing stuff from lua several times |
| 13:51:12 | speachy | that plugin could generate custom clips for tagnavi too |
| 13:51:21 | speachy | and store them under dbtalk/tagnavi/blablabla.talk |
| 13:52:00 | _bilgus_ | the generating the clips might be a bit too far unless its a rbutil kinda thing |
| 13:52:07 | speachy | I still think that it would be good to extract "spelling letters" from the LANG file and into standalone clips is a good idea, so that way we could support non-english letters |
| 13:52:18 | _bilgus_ | I will eventually get back to trying to get espeak running |
| 13:52:37 | speachy | rbutil (and the cmdline stuff) would need to consume the same data files |
| 13:53:05 | _bilgus_ | oh you mean the actual text prompts then? |
| 13:53:38 | speachy | yeah, everything. we ship stuff for the built-in tagnavi, but this way we can support voicing custom entries |
| 13:54:06 | speachy | ie for string X, look it up in our LANG, if it's not found then look on-disk |
| 13:54:36 | _bilgus_ | with sensible naming it wouldn't be too bad |
| 13:55:11 | _bilgus_ | everything sits in its folder and a simple fileexists() so you can leverage the cache |
| 13:55:17 | speachy | granted with tens of thousands of tracks we might have to worry about performance simply looking for the correct clip |
| 13:56:05 | _bilgus_ | I think you'd be fine might have to bump he cache a bit or some memoizing |
| 13:56:40 | _bilgus_ | its not that hot till you start doing it for every track |
| 13:57:19 | _bilgus_ | but figure you only voice 10 of the 100 loaded |
| 13:58:43 | speachy | do we cache clips? |
| 13:58:54 | speachy | (ie not in the main voice db) |
| 14:00 |
| 14:00:06 | _bilgus_ | I think so |
| 14:00:29 | _bilgus_ | but I was saying the directory cache will buy most of that |
| 14:00:43 | speachy | I want to prototype maknig the letters into standalone clips |
| 14:01:23 | speachy | ...and having each language generate its complete alphabet. |
| 14:01:44 | _bilgus_ | similar to invalid voice |
| 14:01:51 | speachy | yeah |
| 14:01:59 | speachy | but fallbacks are messier |
| 14:02:04 | _bilgus_ | not a bad idea it should be easy to get it to use them |
| 14:02:22 | speachy | because it's really common to have files with letters not part of the active language |
| 14:02:27 | _bilgus_ | going to be a bit around that clip lookup part |
| 14:02:50 | speachy | .rockbox/lang/letters/<codepoint>.talk |
| 14:03:18 | speachy | problem is that (eg) 'A' could be voiced very different by oh, most latin-based languages. |
| 14:03:29 | speachy | but how do you implement a sane fallback? |
| 14:03:48 | _bilgus_ | codepoint 124? |
| 14:04:20 | _bilgus_ | sane but not very human friendly lol |
| 14:04:28 | speachy | eg if you're english, how does 'Akçakoca' get spelled? |
| 14:05:06 | speachy | (that's Turkish, pronounced ah-ka-cha-ko-ja) |
| 14:06:10 | speachy | eg is the turkish 'c' is the same codepoint as the english 'c' ? |
| 14:06:38 | _bilgus_ | A k c(cedilla) a k o c a |
| 14:06:58 | _bilgus_ | or c(codepoint 123) |
| 14:07:02 | speachy | (yes, apparently) |
| 14:07:20 | speachy | so spelling is going to be necessarily naive since we don't know what language the originating string is from |
| 14:07:40 | speachy | for non-native strings I mean |
| 14:07:48 | _bilgus_ | only so much we can do given our limited size and processing |
| 14:08:24 | _bilgus_ | the ones with it to spare don't even get it completely right |
| 14:08:24 | speachy | so it's easy to extend the current approach to different languages (.rockbox/lang/<lang>/<codepoint>.talk) |
| 14:09:16 | speachy | but maybe the fallback should simply be the numeric "codepoint 123" until we can figure out something better.. |
| 14:09:40 | _bilgus_ | its no so gret but it is 'Right' |
| 14:09:45 | _bilgus_ | great |
| 14:09:55 | speachy | and it's a lot better than our current engrish-only approach |
| 14:10:11 | _bilgus_ | less confusing in the end |
| 14:10:48 | _bilgus_ | that alos would atleast give you a smaller set of things to keep in sync |
| 14:11:26 | speachy | we'd keep numeric stuff in the core voice file |
| 14:11:27 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 14:11:36 | speachy | and significant punctuation |
| 14:11:46 | speachy | (eg %, /, etc) |
| 14:13:08 | speachy | two approaches to generating the alphabet −− require the alphabet explicitly in a string, or.. take all of the strings in the translation and extract the full set of codepoints used |
| 14:13:25 | _bilgus_ | I don't think it'll be any worse than custom talk clips processing wie |
| 14:13:26 | speachy | the latter approach would be useful for CJK |
| 14:13:43 | speachy | (ie where there is no "alphabet") |
| 14:14:12 | _bilgus_ | bbl |
| 14:14:20 | speachy | kk |
| 14:15:54 | | Quit dconrad (Ping timeout: 252 seconds) |
| 14:27:52 | *** | Saving seen data "./dancer.seen" |
| 14:40:56 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 14:45:45 | | Quit dconrad (Ping timeout: 276 seconds) |
| 14:54:20 | | Join lebellium [0] (~lebellium@2a01cb0405d07f0098ddb854a2219d13.ipv6.abo.wanadoo.fr) |
| 15:00 |
| 15:11:48 | user890104 | if you don't omit the kanjis, things are going to get out of control |
| 15:12:03 | user890104 | but then it won't support chinese |
| 15:13:06 | speachy | Voicing CJK codepoints is complicated; we can't really do an "inclusive" set beyond covering everythign that's in the translation already |
| 15:13:21 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 15:14:00 | speachy | (the question about how to fall back is a challenge; I suppose it's probably safe to fall back to english unconditionally..) |
| 15:14:51 | speachy | it's an ugly problem that we don't have the on-device resources to solve properly |
| 15:18:15 | | Quit dconrad (Ping timeout: 260 seconds) |
| 15:42:23 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 15:46:49 | | Quit dconrad (Ping timeout: 252 seconds) |
| 16:00 |
| 16:14:48 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 16:19:37 | | Quit dconrad (Ping timeout: 268 seconds) |
| 16:27:53 | *** | Saving seen data "./dancer.seen" |
| 16:37:23 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 16:41:31 | | Quit dconrad (Ping timeout: 244 seconds) |
| 16:58:33 | | Join PheralSparky [0] (~S|h|a|w|n@user/shawn/x-4432647) |
| 17:00 |
| 17:03:54 | | Quit qf (Ping timeout: 246 seconds) |
| 17:13:57 | | Join Moriar [0] (~moriar@107-200-193-159.lightspeed.stlsmo.sbcglobal.net) |
| 17:19:53 | | Join qf [0] (~qf@user/qf) |
| 17:25:48 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 17:32:26 | | Quit dconrad (Ping timeout: 252 seconds) |
| 17:55:55 | | Quit othello7 (Ping timeout: 244 seconds) |
| 18:00 |
| 18:00:10 | winsim-buildbot | New Windows simulator build round started. Revision 801260dd79, internals: VIRT_PTR needs to point at an *invalid* unmapped address by Solomon Peachy |
| 18:01:03 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 18:03:16 | | Quit lebellium (Quit: Leaving) |
| 18:05:25 | | Quit dconrad (Ping timeout: 248 seconds) |
| 18:22:03 | | Quit hactar|ant (Read error: Connection reset by peer) |
| 18:23:20 | | Join hactar|ant [0] (~zem@97-115-76-84.ptld.qwest.net) |
| 18:26:18 | | Join qf1 [0] (~qf@user/qf) |
| 18:26:20 | | Quit qf (Ping timeout: 252 seconds) |
| 18:27:57 | *** | Saving seen data "./dancer.seen" |
| 18:29:06 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 18:33:38 | | Quit dconrad (Ping timeout: 244 seconds) |
| 18:44:54 | | Join massiveH [0] (~massiveH@2600:4040:a982:5400:acd0:5b7d:cf8c:58e4) |
| 19:00 |
| 19:01:35 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 19:03:43 | | Join qf [0] (~qf@user/qf) |
| 19:04:07 | | Quit qf1 (Ping timeout: 244 seconds) |
| 19:05:56 | | Quit dconrad (Ping timeout: 252 seconds) |
| 19:11:08 | | Join qf1 [0] (~qf@user/qf) |
| 19:11:36 | | Quit qf (Ping timeout: 276 seconds) |
| 19:26:10 | | Quit TorC (Ping timeout: 260 seconds) |
| 19:28:36 | | Join TorC [0] (~Tor@fsf/member/TorC) |
| 19:42:26 | | Quit qf1 (Ping timeout: 248 seconds) |
| 19:44:57 | | Join massive_H [0] (~massiveH@2600:4040:a982:5400:acd0:5b7d:cf8c:58e4) |
| 19:46:29 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 19:47:45 | | Quit massiveH (Ping timeout: 248 seconds) |
| 19:51:08 | | Quit dconrad (Ping timeout: 268 seconds) |
| 20:00 |
| 20:28:00 | *** | Saving seen data "./dancer.seen" |
| 20:31:12 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 20:35:46 | | Quit dconrad (Ping timeout: 252 seconds) |
| 21:00 |
| 21:07:13 | aaabbb | speachy: so the opus optimizations that put key datastructures in tcm are back then, for arm targets? that is good. there is not much room for optimizing opus decoding on arm targets, then? (other than optimizations that may be done by upstream libopus) |
| 21:07:55 | aaabbb | last time you had said they were removed as the lifetimes of the objects, previously put in tcm, had changed due to upstream libopus changing, and so they were moved out of tcm. they are back in? |
| 21:31:55 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 21:36:50 | | Quit dconrad (Ping timeout: 260 seconds) |
| 21:53:40 | | Join othello7 [0] (~Thunderbi@pool-100-36-176-164.washdc.fios.verizon.net) |
| 22:00 |
| 22:04:40 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 22:08:54 | | Quit dconrad (Ping timeout: 252 seconds) |
| 22:26:10 | | Quit Moriar (Ping timeout: 248 seconds) |
| 22:28:04 | *** | Saving seen data "./dancer.seen" |
| 22:48:59 | _bilgus_ | aaabbb, I guess the better question at this point is what are you hoping for? |
| 22:49:35 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 22:49:38 | _bilgus_ | it appear 256 is around 70-90mhz if you need better I'd suggest mp3 |
| 22:50:40 | _bilgus_ | we have a codec test plugin too |
| 22:54:14 | | Quit dconrad (Ping timeout: 272 seconds) |
| 23:00 |
| 23:34:15 | | Join dconrad [0] (~dconrad@152.117.104.232) |
| 23:40:33 | | Quit dconrad (Ping timeout: 268 seconds) |
icon identifies that the person is a core developer (has commit access).