=== _salem is now known as salem_ === salem_ is now known as _salem === IdleOne- is now known as IdleOne === Guest83203 is now known as Tm_T === chihchun_afk is now known as chihchun === ara is now known as Guest5294 [07:15] good morning [07:26] davmor2, yo [08:25] hey all [08:25] new to ubuntu touch [08:27] Is private Internet access available on touch then? [08:39] Maple: you mean vpn? [08:40] yes vpn [08:41] Maple: it's not hooked up to the UI yet, but I suspect it will be in the future. [08:42] Is possible to set up from terminal? [08:57] Maple: not sure, not tried [08:58] private internet access is the sponsor here? [08:59] Maple: Here? This is an irc channel, it doesn't have sponsors... [08:59] (other than the people who maintain it [freenode / pdpc] and Canonical who sponsor Ubuntu development) [09:01] Just saw something about them being a sponsor on the login page [09:01] I think lol [09:02] Maple: login page of what? [09:04] Maple: you're using a web page to view this irc channel? I'm interested to know where that page is. [09:08] lost connection there for a min [09:08] I think they sponsor [09:08] freenode [09:09] cwayne: Wot not wall [09:11] Maple: so they do! [09:14] They should have a channel I spose [09:15] ill try to get hold of em === chihchun is now known as chihchun_afk === datkinso1 is now known as datkinson [09:25] bzoltan, so https://code.launchpad.net/~ubuntu-core-dev/ubuntu-seeds/ubuntu-touch.vivid needs an upload? === chihchun_afk is now known as chihchun [09:25] dholbach: yes, the ubuntu-sdk-qmake-extras needs to be pulled by the ubuntu-sdk-libs-tool in all 15.04 chroots in order to enable the the new templates [09:26] dholbach: usually Mirv does it for the SDK, but he is still out ... and all other regular folks are covered with other duties. [09:26] bzoltan, I don't know what the latest is in terms of seed changes - touch grew a bunch of processes on their own [09:26] ogra_, ^ do you know what to do? [09:27] dholbach, yes, but i dont have time right now (i told bzoltan in ~2h ... but seems he cant wait) [09:27] ogra_: 2 hours I can wait :) But I waited 5 days already [09:27] you need to merge the seed change and rebuild the meta package after you ran ./update in the source package dir (and adjust the changelog afterwards) [09:28] ogra_, right, that I know - I was just wondering if there was any more red tape... [09:28] right, and i told you after two days that every core-dev can do it ... you could just have asked in #ubuntu-devel for a sponsor [09:28] ogra_, so it can just be sponsored like anything else? [09:29] yeah [09:29] its just a universe package [09:29] (seeds are owned by core-dev though) [09:29] right [09:29] I'll take a look [09:29] thanks [09:30] davmor2, u wot m8 [09:31] cwayne: WOT NO WALL [09:31] Thank you ogra_ and dholbach. [09:31] davmor2, so was wondering what happened re: the online accounts being broken stuff [09:31] was the signon-apparmor-extension reverted? [09:32] Good morning all; happy Monday, and happy Australia Day! :-D [09:40] bzoltan, uploaded [09:40] dholbach: that was fast :D [09:44] zbenjamin: I just have flashed a fresh RTM image on my krillin without dev mode... I have no problem adb shelling in when the screen is locked [09:44] bzoltan: ok, i wonder what that change is then [09:44] depends how you flashed [09:44] zbenjamin: I am not going to ask it :) I am happy with the present situation [09:44] --developer-mode disables screen lock checks [09:45] if you flash normally it stays on [09:45] ogra_: I have not used that switch [09:45] weird [09:45] ogra_: i just flashed the 213 [09:45] then you shouldnt get in [09:46] ogra_: the adb was off at first of course, I turned the dev mode on in settings, set the passcode [09:46] a normal flash wont put the override in place so you should not be able to do the initial connection if you have a locked screen [09:46] ogra_: but after that plug out, plug in ... screen locked and I can adb shell in. But I am extremely happy with that [09:47] well, that would be a release blocker [09:47] bzoltan, we are talking abour rtm, right ? [09:47] ogra_: the adb is not on by default... it comes alive after I enabled the dev mode [09:48] 2015/01/26 11:30:23 Flashing version 213 from ubuntu-touch/ubuntu-rtm/14.09-proposed channel and server https://system-image.ubuntu.com to device krillin [09:48] has nothing to do with on by default === chihchun is now known as chihchun_afk [09:50] sergiusens, u-d-f only puts the adb_onlock file in place when developer-mode is set, right ? [09:51] ogra_: correct; I can double check, give me a sec [09:51] bzoltan, did you use --bootstrap ? [09:51] ogra_: no [09:51] ogra_: should I? [09:51] ah, and you flashed that device before with --developer-mode at some point i guess [09:51] only on clean installs as well [09:52] cwayne: it was http://i1.ytimg.com/vi/BPeDzRW9NIQ/maxresdefault.jpg [09:53] ogra_: sure I did [09:53] ogra_: ok, so developer-mode enables the two files, adb_onlock and the developer_mode file [09:53] ogra_: and password requires a 'format data' [09:54] ogra_: all these rules become moot once we ship with a non adbable recovery [09:54] ogra_: we can just put those file in the recovery image [09:55] ogra_: since it's not even 7AM here, I'll bbl :-P [09:55] sergiusens, right, i'm just worried we leave that file behind somehow on fresh flashes [09:55] i think we should perhaps forcefully delete it if dev-mode isnt set during flash [09:55] i think it isnt even deleted when using --wipe [09:55] because it lives one dir level up from what gets wiped [09:56] ogra_: I used the wipe [09:56] bzoltan, right [09:57] but i bet you have a /userdata/.adb_onlock file [09:58] even after --wipe [09:59] ogra_: sergiusens: I have two concerns: 1) It will be seriously disturbing for app developers that they are forced to manually unlock the screen whenever they want to run their app from the SDK. But i do understand that it is not a priority :) 2) i just want to run my UITK tests without manual hustle ... and it does involve reboots, so I need to be able to flash the device just as i do now [10:00] bzoltan, 1) use --developer-mode when flashing [10:00] ogra_: yes i do have the .adb_onlock [10:00] 2) see 1) [10:00] 3) put the file in place yourself by touching /data/.adb_onlock from recovery [10:00] ogra_: 2) yes, that is what i use... I hope that behavior will not change. [10:00] no [10:01] the point is that if you later flash without --developer-mode i think the file needs to be wiped [10:02] ogra_: yes, that sounds logical [10:02] developers should keep that flag ... if they re-flash to give the phone to someone else and dont use that flag, the file should be wiped so the screen lock check is there for the new user [10:02] though i guess if you flash to give it away you also use --bootstrap to reformat the partitions [10:03] ogra_: So we expect the developers to jump thru loops and do some magic in order to enable them creating super simple apps? :) [10:04] bzoltan, ask the security team ... that stuff isnt my desgn, i only implemented it ... [10:04] long term target is key auth [10:04] ogra_: I know :) I do not shoot the messenger [10:04] which means you will need to approve a key after each flash [10:04] on screen [10:05] thats not much different than unlocking once [10:05] ogra_: and our short term target is to engage ... making such a silly policy is not exactly doing that [10:05] (you wont have to do it subsequently though ) [10:06] the point is that someone stealing your phone shuldnt be able to adb in when he doesnt know the PW/PIN [10:06] even if you switched dev mode on [10:06] (in the UI that is) [10:07] if you have overridden that at flash time already, we assume you are a developer and leave the feature off completely [10:07] (i.e. if you use --developer-mode) [10:07] ogra_: that is fine.. but under a single connection I would expect the device to be available ... so it is OK to request pin code once I plug in... but not every minute [10:08] we onyl request it on connect events, if you stay connected you wont have any new unlock request [10:09] anyway, you'Re a developer, use --developer-mode ... it is for you :P [10:10] ogra_: hmm.. I am a developer, but I do not wish to flash ... [10:17] bzoltan, so just touch the file then [10:22] ogra_: i use --developer-mode and --password=phablet to flash devel image but i still can't sudo with password. any idea? [10:23] shuduo, does unlocking the screen with that pasword work ? [10:24] ogra_: the unlock pin is set with another 4 digital number [10:24] well, that means you did set a new password then ... in the wizard i guess :) [10:24] use the pin with sudo [10:25] ogra_: great. it works now. thanks! [10:25] --password=phablet will only work if you keep it set and dont change anything via the UI ... then whatever you changed to will be used === chihchun_afk is now known as chihchun === chihchun is now known as chihchun_afk [10:47] oSoMoN, o/ [10:47] rpadovani, hey! [10:48] oSoMoN, I have a couple of questions about https://code.launchpad.net/~rpadovani/webbrowser-app/addressBarFullWidth/+merge/239039/comments/612176 [10:49] oSoMoN, to fix the module problem I have to add it to debian/control, right? [10:51] rpadovani, yes [10:52] oSoMoN, cool, and what about other two errors? I thought eventually and equals were autopilot valid commands [10:52] rpadovani, they are, but you need to import them to use them [10:54] oSoMoN, oh, I see - thanks! I also see you reply to all my branches, I'll take care to fix them asap :-) [10:54] thanks again! [10:54] cheers! === ecloud_ is now known as ecloud === chihchun_afk is now known as chihchun [11:05] oSoMoN, sorry to bother again, in debian/control I have to add qtdeclarative5-ubuntu-keyboard-extensions0.1 in Source: webbrowser-app/Section: x11 or Package: webbrowser-app/Architecture: any or both? Or others? [11:07] rpadovani, I think you’ll have to add it to both Source and "Package: webbrowser-app", because when building the package the unit tests are run, and AddressBar.qml is unit-tested [11:08] thanks! Just pushed, let see if now works [11:19] davmor2: hi! When you found bug 1413655, did you have only the U1 account, or others too? [11:19] bug 1413655 in ubuntuone-credentials (Ubuntu) " --switch removes accounts" [Undecided,New] https://launchpad.net/bugs/1413655 [11:20] mardy_: all of them [11:20] bar flickr I think [11:22] davmor2: you mean you had U1 and flickr? [11:23] mardy_: no I was signed into all of them except flickr [11:25] hmm, my flo running 171 ubuntu-touch/ubuntu-rtm/14.09-proposed refuses to connect to any wireless networks... [11:26] haha, now it works. Stupid tablet. [11:26] bzoltan, I think the fix has landed already [11:27] dholbach: Cool, thanks [11:42] popey: it hates you you just have to hate it back harderererer === _salem is now known as salem_ === chihchun is now known as chihchun_afk === chihchun_afk is now known as chihchun [12:05] bzoltan: the concerns you have you have to take to ogra_ and the security team, not me; I am against all these "features" in the first place (the files to flag access) [12:05] heh === chihchun is now known as chihchun_afk [12:06] sergiusens: ogra_: i tried to talk to people before, but it seems to be a solid requirement [12:06] well, the key support will eventually fix all this === chihchun_afk is now known as chihchun === JMulholland_ is now known as JMulholland === dandrader is now known as dandrader|afk === rickspencer3_ is now known as rickspencer3 === chihchun is now known as chihchun_afk [12:57] Hi all! [12:58] I tried ubuntu touch a couple of months ago and some bugs were really annoying what's about today?? === MacSlow is now known as MacSlow|lunch === dandrader|afk is now known as dandrader === alan_g is now known as alan_g|lunch === MacSlow|lunch is now known as MacSlow === alan_g|lunch is now known as alan_g [14:33] mandel, did you see my comment on they check-hash bug report? I confirmed our DownloadTracker isn't getting the error signal from udm, i haven't been able to confirm if udm emitted it though === lool- is now known as lool [14:34] kenvandine, I missed it, sorry, again in a meeting (gosh) but I'll take a look asap [14:34] kenvandine, AFAIK udm does emit it or else the tests would not pass when building udm :) [14:35] if so, then we aren't getting it [14:35] mandel, we have a warning if it fails to connect to the signal, which isn't getting logged [14:35] and i added some debug output in our callback, nothing [14:35] kenvandine, I'll take a look, no need to worry, I should be able to find the cause really fast [14:36] mandel, thx [14:42] mardy_: so, any idea how to fix it so that unconfined click packages actually used "unconfined" as the profile instead of $APPID? [14:43] jgdx, i'm anxious to see CI results for your 2 AP fix branches, but I noticed that jenkins isn't subscribed to the review... [14:44] kenvandine: jenkins usually auto-adds itself as a reviewer to any MPs for branches it watches [14:46] dobey, i know... [14:46] but it didn't for 2 MPs jgdx filed this morning [14:46] which is odd... [14:46] oh... just now it added it... way late :) [14:47] kenvandine: sometimes it's quite slow :) [14:47] doesn't usually take this long :) [14:47] ogra_: the last problem is that /userdata/.adb_onlock needs to be created as root :( [14:48] bzoltan, thats the reason why u-d-f does it for you :) [14:49] ogra_: I mean, I do not want to flash the device just for that [14:49] bzoltan: that's a private interface, I wouldn't rely on it for functionality [14:50] sergiusens: ogra_: we need an interface to allow the developer to set that flag [14:50] bzoltan: ask the product team and it will make it into our sprint planning [14:50] dobey: I just sent an email about it === Guest99704 is now known as balloons [14:50] dobey: I think that the immediate fix is the one I suggested; let's see if jdstrand_ can play some magics later on === balloons is now known as Guest37568 [14:51] bzoltan, this intertface is going away once we switch to key auth [14:51] not sure it is worth spendin any time on it === Guest37568 is now known as balloons_ [14:51] developers that want to use automation should od the initial flash of the device with --developer-mode [14:52] kenvandine, yeah, me too [14:52] jgdx, there's a merge conflict in your security-privacy-add-sim-unlock-1378883 branch [14:52] kenvandine, right [14:52] ogra_: I could put a simple switch in the SDK UI so the developers could create that file [14:52] please dont [14:52] bzoltan: that's making it easy to enable a security hole [14:52] that would be a horrid hack and forcing a securioty hole open [14:53] ogra_: I do not :) because I can not [14:53] good [14:53] ogra_: not bigger than telling the developer to flash the device with --developer-mode [14:53] juts tell devs they should use --developer-mode on first flash of the device [14:53] if you dont --wipe or --bootstrap then the file will stay [14:53] ogra_: do we tell the developers that it will open the adb connection wide up? [14:54] we didnt tell the developers at all yet [14:54] since this interface is actually only for automation [14:54] ogra_: but the u-d-f --help lists the --developer-mode [14:54] * ogra_ was actually trying to get the options removed from --help and the manpage ... but i was overruled [14:55] ogra_: I know it is for CI and for me :) [14:56] ogra_: all right, it is fine ... the SDK will tell the developer to unlock the screen each time the unlocked device is pluged in ... after it is unlocked the SDK does all the magic via ssh, so i do not care about the adb anymore ... [14:57] right [15:11] mardy_: you forgot to add jdstrand_ to cc: list, so i replied to add him and ask his thoughts. hopefully he'll be on soon and have a good suggestion about how to fix it. [15:12] * jdstrand_ is here now [15:12] give me a couple minutes === jdstrand_ is now known as jdstrand [15:12] * jdstrand is helping with another phone bug atm [15:14] kenvandine, fixxd [15:14] jgdx, thx [15:14] kenvandine, this is also a deb run, but feel free to review the sim unlock [15:14] jgdx, will do, once CI runs for debs [15:14] kenvandine, ack [15:15] jdstrand: sure, no problem. thanks [15:20] dobey: eh, thunderbird should also warn me when I mention people but forget to add them to CC ;-) === mardy_ is now known as mardy [15:23] ricmm, hey, seems like your recent platform-api update to vivid made unity8-desktop/amd64 stop starting [15:24] Saviq, greyback_, ^ moved there since he's not on -unity [15:24] it worked for me, but I did have a qtmir built myself, so might be a rebuild is all that's needed? [15:25] * Saviq tries again after having reverted to distro qtmir [15:25] seb128: whats the error? [15:26] ricmm, dunno, get a blank screen after updating ubuntu-application-api2-desktop [15:26] hmm still working here [15:26] ricmm, unity8.log has [15:26] "Ubuntu Platform API: Unable to load selected module. -- Aborting" [15:26] https://code.launchpad.net/~gerboland/platform-api/add-dlerror/+merge/241792 <- this would help for next time [15:27] ricmm, http://paste.ubuntu.com/9883282 [15:27] that's the unity8.log [15:33] jgdx, what is this connectivity binding used for? Is it the wrapper for NM? [15:34] kenvandine, no, it's a wrapper for the private connectivity api trusted apps can use [15:34] jgdx, so we can't use the Ubuntu.Connectivity CPI? [15:34] s/CPI/API [15:35] kenvandine, isn't that the public one without sim/modem unlock features? [15:35] i really don't know :) [15:36] just making sure we aren't re-inventing :) [15:36] or... rather had something old that isn't needed anymore [15:36] jgdx, i wasn't sure what that was used for in cellular and wifi [15:37] kenvandine, it's brand spankin' new, 'published' here: https://bugs.launchpad.net/indicator-network/+bug/1374082/comments/7 [15:37] Launchpad bug 1374082 in indicator-network (Ubuntu Utopic) "no API to unlock a specific sim" [Undecided,Triaged] [15:37] seb128: whats UBUNTU_PLATFORM_API_BACKEND set to? [15:37] do you have 2.7.0 of -desktop and -test ? [15:37] kenvandine, we need this in the cellular panel as well, so eventually we should move it into src/ ? https://bugs.launchpad.net/ubuntu/+source/ubuntu-system-settings/+bug/1414670 [15:37] Launchpad bug 1414670 in ubuntu-system-settings (Ubuntu) "Consolidate connectivity bindings from security-privacy and cellular" [Wishlist,In progress] [15:38] ricmm, how do I check that env variable if unity8 fails to start? [15:39] ricmm, -test is not installed [15:39] jgdx, ok, the backend for Ubuntu.Connectivity is provided by indicator-network, but i guess that private API probably isn't exposed [15:39] jgdx, so carry on :) [15:39] jgdx, thanks for explaining [15:39] kenvandine, that's my understanding. :) [15:39] np [15:40] ricmm, works after installing -test, missing depends? [15:40] Saviq, ^ might explain why it works for you [15:40] seb128: ricmm: just a note, qtmir is setting UBUNTU_PLATFORM_API_BACKEND to either desktop_mirclient or touch_mirclient, if it hasn't been set already. It decides which based on if DESKTOP_SESSION contains the string "mir" === dandrader is now known as dandrader|lunch [15:40] seb128: it was just a guess, but thats odd [15:40] lemme think about it [15:42] ricmm, christownsend confirms that installing -test makes it work again for him as well [15:44] seb128: I see whats going on, its my bug/bad [15:44] :) [15:44] good [15:45] I'll fix in a sec [15:45] chris is opening a bug report [15:45] thanks [15:46] ask him to add me to it and ping m [15:46] me [15:46] bfiller, I'm seeing this bug https://bugs.launchpad.net/camera-app/+bug/1414682 in the camera app, just flashed my krillin and updated the camera app [15:46] Launchpad bug 1414682 in camera-app "Controls disappear after taking photo" [Undecided,New] [15:49] ricmm: Hey, I'll enter a bug report against platform-api about desktop breaking. [15:50] ChrisTownsend: thanks, assign me to it [15:51] ricmm: Ok, will do. Ping me if you want more info or whatever in the bug report. [15:51] jgdx: is your disk full? also can you do a "click list | grep camera" [15:51] bfiller, "false alarm", I' [15:51] bfiller, I deleted my Pictures folder [15:51] ChrisTownsend: no I already know whats going on, bug is just to track [15:51] and close [15:51] updating bug… [15:51] jgdx: whew [15:51] :) [15:51] :D [15:51] indeed [15:52] ricmm: Sweet! [15:52] jgdx: there is a know but that the controls disapear when switching between front and back cameras until you tap on the viewfinder [15:52] bug [15:54] ricmm: FYI, https://bugs.launchpad.net/platform-api/+bug/1414692 [15:54] Launchpad bug 1414692 in platform-api "Latest update to platform-api breaks Unity 8 Desktop" [Undecided,New] [15:55] bfiller, yeah, I see it. [15:56] cwayne, nhaines: either of you have problems with something hard freezing on n5? past couple of days i've had many cases where the screen wouldn't turn back on, and i couldn't phablet-shell into the device either, and had to reboot [15:58] ChrisTownsend: thanks [16:26] kenvandine, jgdx, hey, I'm looking at bug #1412523 ... do you have a preference on how the "enable button" logic should be handled? between having the button using a "enabled: " or having the different entry widgets doing things on onTextChanged? [16:26] bug 1412523 in ubuntu-system-settings (Ubuntu) "In "Switch to passcode", "Set" is available when passcode or passphrase not supplied" [Low,In progress] https://launchpad.net/bugs/1412523 [16:26] like either we do [16:27] enabled: oldmode=swipe && (if passcode, length = 4, if passphrase, length >0) || oldmode=passcode ... [16:29] seb128, i'm fine with the complex list of use cases [16:29] kenvandine, great, thanks [16:29] np [16:40] seb128, the data that is used on the battery graph, is that stored in a file somewhere by any chance? [16:41] pmcgowan, it comes from upower and the data are in /var/lib/upower [16:41] seb128, thanks [16:41] seb128, i'm landing your focus branch now... just waiting for the ppc build in the silo :/ [16:41] kenvandine, great, thanks === VargaD_ is now known as VargaD === dandrader|lunch is now known as dandrader [17:03] bfiller: not sure who will do the camera-app upload for bug #1414625, but can you have whoever does ping me when they do, and I'll add a comment to the store review so future reviewers will know what is going on? [17:03] bug 1414625 in apparmor-easyprof-ubuntu (Ubuntu) "Temporarily grant additional rights to camera-app so it can read free disk space" [Undecided,New] https://launchpad.net/bugs/1414625 [17:03] popey: fyi, ^ if you see that, ping me-- adding a read_path to camera-app [17:04] seb128, is there any tool to display those files? i assume the first item is a time stamp [17:04] pmcgowan, no tool that I know, sorry [17:04] seb128, k [17:05] pmcgowan, yeah, it's timestamp, %charge, status [17:05] pmcgowan, there is a new data point written on changes [17:05] jdstrand: I'll do the camera-app upload [17:05] jdstrand: and will ping you [17:05] so if you work plugged at 100% charge for 3 days there are no data during this time [17:05] seb128, makes sense [17:05] pmcgowan, are you looking at a specific issue? [17:05] trying to track down some precipitous drops in battery level [17:06] like 30% to 0 in minutes [17:06] jdstrand: assuming will need changes to camera-app json and a new apparmor-easyprof package? [17:07] Elleo: just a heads up, thanks to a few nice/dirty tricks to the Brute Force Matcher in skeyer it's query time is down to around 200ms for a long word that was taking 5 seconds to compute before... [17:07] The word is the swipeHint of the word beautiful.. Which is around 40 characters [17:08] saidinesh5: nice :) [17:08] And for normal words the response time is around 27ms [17:08] sweet [17:08] So i think we've found your hunspell replacer if not anything :p [17:08] heh [17:09] bfiller: just camera-app [17:09] bfiller: I haven't updated the bug yet [17:09] (This one is kind of aware of key board layouts too) [17:09] jdstrand: if I see it in the store? [17:09] saidinesh5: cool [17:09] popey: yes [17:09] jdstrand: cool [17:09] jdstrand: ok [17:09] popey: the camera-app already triggers a manual review becuase of video_files and picture_files policy groups [17:09] Now will have to wait till i get my laptop so that i can test this out on my nexus 7 [17:10] With Ubuntu touch [17:10] popey: something new is being added, and I wanted to comment on it for posterity [17:10] Right ho. [17:11] Also need to optimize the memory requirements of this... [17:11] saidinesh5: let me know how it works out on the device :) [17:11] Yup [17:11] Oh or i can compile it on the device itself now.... Hold on [17:12] saidinesh5: what sort of tricks are you doing for those speed ups? [17:12] 3 little tricks mainly [17:13] 1) early return in edit distance method if you're trying to find only k differences... [17:13] 2) if you're trying to find only 5 words, adjust k accordingly based on the results you've collected so far [17:13] 3) if manhattanDistance(input,word) > k don't bother going for editDistance [17:14] And then if word lengths differ by more than k, don't bother computing manhattanDistance either [17:14] ah, interesting [17:14] Manhattand distance is basically the number of times each character occurs [17:14] In the string [17:15] Difference of that in the two strings i mean [17:15] yeah [17:15] what sort of dictionary size are you working with? [17:15] So that's O(n) [17:16] Android dictionary... Around 180000 words maybe? [17:16] interesting, hunspell's US dictionary is only about 62k entries [17:16] But to search for a word I'm only searching the words starting with the neighbors of a key [17:17] So my search space is probably the same sized [17:17] ah, right [17:17] 9 neighbors of a key assuming a grid of keys [17:17] so there's an assumption that the user is correctly spelling the first character (but possibly missing the key) [17:17] Yeah [17:18] Otherwise the results are too noisy [17:18] that could fall down on some misspellings like "psychologist" -> "sychologist" [17:18] That's the next todo item actually :p [17:19] ah, cool :) [17:19] Add a frequently misspelled list of words basically [17:19] right, seems reasonable [17:19] Mhm [17:19] but could be a fair amount of work compiling those lists for all languages [17:20] But otherwise, the swipe for psychologist and sychologist is quite different [17:20] yeah [17:20] I'm sure Google open sourced some words [17:20] Word lists i mean [17:20] if you do find some existing word lists like that could you point them my way too, would be handy to have a look at [17:21] Oh also totally forgot, for normal words we can search the whole word list btq [17:21] Normal words = ~ 8 characters long [17:22] ah, right; so these optimisations only get applied to long slow words? [17:22] More like only when the user swipes a word as opposed to types a word [17:23] But yeah will have to look into that change... [17:23] You have access to the sources right? [17:23] Lib/bruteforcematcher.cpp [17:24] saidinesh5: cool, will have a look [17:24] Also will probably test out simstring library after this [17:24] But need more exhaustive tests first [17:25] Need to find out the average precision and all [17:25] interesting, hadn't seen simstring before [17:25] just reading up on it now [17:25] Looks interesting actually it had it's own binary format and [17:26] Mmaps it... [17:26] will have to have a play with that myself at some point [17:27] Also do you know any way to mmap a QList [17:27] Can't seem to find any documentation / sample code [17:28] Oh grr no adb on this machine [17:28] And can't ssh without my public key on the device [17:28] nope, not something I've tried before, not sure if the Qt internals would lend themselves to that sort of treatment well (without risking issues between Qt versions) [17:29] Well there is a QFileDevice:map [17:29] For calling mmap [17:29] It returns a uchar* [17:30] The thing is QString seems to take care of its own memory allocation internally [17:30] So not sure what happens when i a QByteArray [17:31] Do a qdatastream >> qstring i mean [17:31] yeah, I'd suspect it'd end up effectively doing a read of all the data, so you'd end up with it all in memory anyway [17:31] that's just a guess though [17:40] Not sure for me there is 2x the string data [17:41]