| <<<Back 1 day (to 2015/08/16) | 20150817 |
kens | chrisl ping | 08:05.02 |
chrisl | kens: pong | 08:05.30 |
kens | Marcos opened a new bug #696149 against UFST with a change I made | 08:06.00 |
| Can you tell me how to get and build a UFST version of GS ? | 08:06.12 |
chrisl | You need to get UFST from subversion | 08:06.37 |
kens | Hmm, OK well I can probably do that, I think I still have a SVN client installed, let me see | 08:06.55 |
| Actually no I don't, so I'll start by getting that | 08:07.23 |
chrisl | I can zip up ufst if that would be easier..... | 08:07.41 |
kens | It probably would be, yes | 08:07.50 |
chrisl | Hmm, let me trim some of of the not needed stuff - otherwise it's 305Mb..... | 08:09.15 |
| zipped! | 08:09.20 |
kens | Crikey.... | 08:09.22 |
| I'm picking up a subversion client anyway, just for the future | 08:10.29 |
chrisl | Okay, in my home dir on casper is ufst-6.x.zip | 08:12.13 |
kens | one moment | 08:12.20 |
chrisl | So, you can use the existing ghostscript-ufst.vcproj if you have the ufst in "c:\ufst" | 08:15.07 |
kens | Hmm, that's not a great directory for me, C:\ is kind of full | 08:15.24 |
| I already have a duplicate of the windows project, I'll modify that | 08:15.43 |
| Just unzippign the file | 08:16.00 |
chrisl | You just need to change where it's got "UFST_ROOT=C:\ufst" on the nmake command line, so it points to wherever you put ufst | 08:16.35 |
kens | Yeah I'm just going to edit hte .vcproj | 08:16.53 |
| Hmm, so how do I access the UFST project ? | 08:18.49 |
| We don't seem to have a target for it | 08:19.03 |
chrisl | Just open the project in VS | 08:19.19 |
kens | The project, not the solution ? OK | 08:19.29 |
| Yeah wants to convert it, just a moment | 08:19.57 |
| OK so it upgrades it to a solutoin | 08:20.41 |
| so far so good, its at least started building | 08:20.59 |
chrisl | No, it creates a solution with the project in it | 08:21.00 |
kens | That's what I mean yes | 08:21.12 |
| Hmm bunch of failures | 08:23.28 |
| Loads of stuff 'already defined' in sstring.obj | 08:23.45 |
| and a whole bunch of other stuff also | 08:24.11 |
chrisl | I haven't tried the VS build for quite a while | 08:24.14 |
kens | Let me clean and rebuild it | 08:24.21 |
| There might have been stuff left lying about | 08:24.32 |
| well differetn stuff is duplicated. | 08:27.14 |
| Strangely they are all of the form 'file.obj: error LNK2005: <symbol> already defined in file.obj' | 08:28.29 |
chrisl | That suggests the obj file is being listed twice in the linker script | 08:29.22 |
kens | That's what I surmised yes | 08:29.30 |
chrisl | Let me dust off my Windows machine and have a try | 08:29.38 |
kens | yeah its including (eg) spprint.obj from debugobj and from ufstdebugobj | 08:31.01 |
chrisl | Er, um..... | 08:31.34 |
kens | THat's just from reading the log, I don't know why yet | 08:32.20 |
| It does appear to be the case ofr all the files with problems though | 08:32.50 |
| I need some coffee, back in a few minutes | 08:33.45 |
| OK so deleting the contents of debugobj leads to a link error 'don't know how to make' on the duplicate files. | 08:41.49 |
chrisl | kens: on the nmake command line, replace "ufst-debug" with "debug" - it should build then, but you'll get the exes in "debugbin" and object files in "debugobj". I'll look at why the ufst specific targets aren't working when I have time | 08:42.06 |
kens | Hmm, OK I don't really mind where they end up right now. | 08:42.28 |
| OK that bujilt, thanks chrisl | 08:48.03 |
chrisl | NP - can't say I've any idea why the ufst target didn't work :-( | 08:48.35 |
kens | It must be something about the way those specific files are built | 08:48.49 |
| Beyond that, no idea | 08:48.55 |
| BTW I may need to ask you to look at this problem anyway, but let me poke it first | 08:49.17 |
chrisl | I dunno - I didn't add those targets, and I've never knowingly used them, so..... | 08:49.30 |
kens | Well I cannot see any difference between the FT and UFST output with teh file Marcos says produces 'bad output' | 09:16.00 |
chrisl | Okay, I'll try it on Linux..... | 09:17.56 |
| Hmm, it ends up with black covering a chunk of the second page - I'm guessing that's not right.... | 09:22.45 |
kens | Not for me on WIndows it doesn't | 09:22.56 |
| Its hard to see how setting the matrix to a tiny value could cause that | 09:23.23 |
chrisl | I'll have a poke at it..... | 09:24.02 |
kens | I'll see if I can get a Linux build running, maybe its jsut that my WIndows build isn't using UFST | 09:24.22 |
Robin_Watts | Helen is just on the phone to our travel agent, talking about some hassles with the car rental we had while on holiday. | 09:25.56 |
| At the same time she's mentioned the December trip. | 09:26.04 |
kens | OK | 09:26.09 |
Robin_Watts | We're looking at going on the 1st, and back on the 7th. | 09:26.20 |
chrisl | Hmm, yes, looking at it the windows UFST build is broken :-( | 09:26.25 |
Robin_Watts | (Helen needs to be back cos she has a concert) | 09:26.38 |
kens | chrisl, well that would explain why I can't see it I guess | 09:26.39 |
Robin_Watts | so it looks like we might be going via Miami. | 09:26.46 |
| Apparently the flights are looking quite full already. | 09:26.58 |
kens | Robin_Watts : Yep, I thnk we were looking at the 30th-6 but I don't recall | 09:26.59 |
Robin_Watts | So we may need to book quite soon. | 09:27.05 |
kens | THat's useful to know, thanks. | 09:27.13 |
| I'll hassle Stella some more :-) | 09:27.19 |
Robin_Watts | kens: That would mean you miss Miles' proposed trip on the sunday. | 09:27.30 |
| But given that mvrhel is in the same position, we might be able to persuade Miles to change the day for the trip. | 09:27.59 |
kens | Robin_Watts : I'd rather miss the trip than fly via Miami | 09:29.09 |
sebras | kens: didn't tor8 find a direct flight..? | 09:29.28 |
kens | sebras, not from the UK, but we cna fly to Nassau and then do an internal flight to Freeport | 09:29.46 |
| Which avoids Miami and US immigration | 09:29.58 |
Robin_Watts | sebras: No direct flights to FPO. There are flights to NAS, then NAS->FPO, but those don't happen every day. | 09:30.14 |
kens | BA do a booking for that trip, but it only flies every 2 days or so | 09:30.20 |
Robin_Watts | hence why kens was looking at coming back on the 6th. | 09:30.31 |
sebras | kens: Robin_Watts: yeah, that's the one. | 09:30.37 |
Robin_Watts | I might sent an email to Miles and mention the issue. | 09:30.46 |
| Possibly he might move the proposed trip date. | 09:30.57 |
chrisl | When is Miles due back from the penury peninsula? | 09:31.16 |
kens | Ths week I htnk, I could be wrong | 09:31.29 |
kens | hopes he wasn't going to Kos | 09:31.46 |
chrisl | Possibly worth dropping him an e-mail about moving the trip | 09:32.06 |
kens | Well I'd like to make a company trip, of course, but it won't worry me too much if we can't | 09:32.34 |
chrisl | I agree, but the earlier we let him know it might be a problem, the better (probably) | 09:33.06 |
Robin_Watts | actually.... I now can't see in his email where he mentions the company trip... | 09:33.19 |
chrisl | Maybe henrys mentioned it on IRC? | 09:33.50 |
kens | one moment, I'm, sure its an email | 09:34.26 |
| mail sent July 21st | 09:34.47 |
| Subject "December staff meeting" | 09:34.56 |
| "It should be beautiful. I expect it will take some time for everyone to figure out what they want to do as far as how long to stay, etc. I will coordinate a company event (with guests) for Sunday. Probably a snorkeling trip." | 09:35.05 |
chrisl | kens: if you also add "UFST_BRIDGE=1 FT_BRIDGE=1" to your nmake command line, that should get you a Windows UFST build (this does not work how I recall it working!) | 09:35.58 |
kens | chrisl on Linux, where's the best place to put the UFST folder ? | 09:36.03 |
| Oh I'll try that, thanks | 09:36.14 |
| Not that I'm using nmake.... | 09:36.21 |
chrisl | Oh yes you are..... | 09:36.30 |
kens | I mean directly | 09:36.35 |
| SO I have to edit the .vcproj | 09:36.42 |
chrisl | You can do it in the GUI as well | 09:37.01 |
kens | Really ? Well that would be easier I guess | 09:37.11 |
| Oh yeah I see | 09:37.25 |
chrisl | It's only more hassle if you want to change it for all the configurations, in which case tweaking the XML is *much* quicker | 09:38.04 |
kens | One will do for now | 09:38.19 |
chrisl | I could swear I changed all this so it was closer to the Unix build :-( | 09:38.55 |
kens | Dunno.... | 09:39.03 |
| chrisl, any pointers on where to put UFST on Linux ? | 09:39.58 |
| My home, the ghostpdl Directory, somewhere else ? | 09:40.18 |
chrisl | anywhere you like | 09:40.36 |
kens | OK, that suggests I need a parameter to say where it is then ? | 09:41.00 |
chrisl | Then you do: ./autogen.sh --with-ufst=<path to ufst> | 09:41.10 |
kens | Got it | 09:41.16 |
| Hmm, first I need to update my checkout :-) | 09:41.39 |
chrisl | That then sets all the stuff that apparently needs set manually on Windows :-( | 09:41.49 |
kens | This'll keep me busy for a while | 09:42.05 |
| Hmm well I guess its using UFST now, it gives me an error 'can't open /Fonts/mtfonts/pclps2/mt3/pclp2_xj.fco' | 09:44.57 |
chrisl | On what platform? | 09:46.43 |
kens | Windows | 09:48.31 |
| My Linux code is so far behind the Git master I htink I'm just going to delete it and clone a new copy | 09:48.59 |
chrisl | There is no distance behind master that should make that necessary | 09:49.33 |
kens | I know, but I regularly break my Linux clones | 09:49.53 |
Robin_Watts | tor8: Hi. Can I get a review of the commits on robin/master please? | 10:29.50 |
tor8 | Robin_Watts: morning. will take a look. | 10:29.58 |
Robin_Watts | I will look at the glut stuff asap. I've got to check something out for fred first though. | 10:30.26 |
tor8 | the signed/unsigned issue in gprf, I've seen warnings about it but what is the range supposed to be? | 10:32.26 |
| if r,g,b are unsigned, then the >255 case will happily take care of overflow | 10:32.53 |
| but then I don't know if they really are supposed to be unsigned or if the r,g,b variable type is wrong | 10:33.14 |
Robin_Watts | tor8: The variables are unsigned. | 10:33.45 |
| They need to be unsigned for the 16*16 muls to work. | 10:34.06 |
tor8 | in that case, why cast to int and compare with < 0? | 10:34.21 |
Robin_Watts | cos the values passed in might not be in range. | 10:34.44 |
| I *could* have the routine take ints, and clip to 0...65535 and then move the values to unsigned for the calculations, but that seems overkill. | 10:35.27 |
tor8 | the input values are also unsigned | 10:36.03 |
| you're using uint32 to do uint16 scaled integer math? | 10:36.44 |
| at least that's what it looks like to me | 10:37.23 |
Robin_Watts | Indeed. | 10:38.26 |
| Yes, the inputs are unsigned. | 10:38.50 |
| That's what I was saying. I *could* make them signed, and clip, and then move them to unsigned for the calculations. | 10:39.15 |
| but that seems a fag. | 10:39.21 |
tor8 | The way the code is now (unsigned everywhere) out of range values would be caught with just the > 255 test | 10:40.03 |
Robin_Watts | tor8: Yes, but they'd clip the wrong way. | 10:40.15 |
tor8 | so why do we need the < 0 test as well? | 10:40.15 |
Robin_Watts | -1 would become max rather than min. | 10:40.33 |
tor8 | right, so the inputs are actually signed int32s? | 10:41.08 |
| scaled so that 0..65535 represent 0..1 | 10:41.49 |
Robin_Watts | The inputs *should* be in range. | 10:42.42 |
| but belt + braces etc. | 10:42.49 |
| but basically, yes. | 10:43.02 |
tor8 | I can't help but think that after all that math, any out of range value could have overflowed to any kind of value | 10:43.13 |
Robin_Watts | Ah! | 10:43.55 |
| This is the post calculation clamping. | 10:44.04 |
| The problem is with the fast + crap calculations. | 10:44.18 |
| r = 65536 - k - c; | 10:44.36 |
| for example | 10:44.41 |
| That can easily give r = -ve numbers. | 10:44.53 |
| and they properly need to be clamped to 0. | 10:45.00 |
tor8 | so if we're relying on overflow behaviour, why not just use int32? *feeling dense today* | 10:46.31 |
Robin_Watts | tor8: The slow but prettier code will fail with int32s. | 10:49.14 |
| because we do (65535*65535)>>16 | 10:49.27 |
| which gives 65535 for uint32s (what we want), but FFFFFFFF for ints. | 10:49.59 |
tor8 | Robin_Watts: right. | 10:57.02 |
| Robin_Watts: all LGTM then. | 10:57.28 |
Robin_Watts | Ta. | 10:57.33 |
| tor8: You've unsettled me enough that I'm tweaking that routine to make the ranges more explicit. | 11:31.49 |
| tor8: OK, tweaked version of that last commit online. | 11:36.30 |
tor8 | Robin_Watts: looks reasonable | 12:13.20 |
Robin_Watts | Ta. | 12:19.51 |
henrys | so white collar criminals in London can't fly direct to the Bahamas... very strange | 13:24.31 |
kens2 | sure we can, to nassau | 13:24.45 |
henrys | ah yes I had my island cities confused ... | 13:27.09 |
Robin_Watts | hopes that email didn't appear ungrateful. | 13:52.09 |
| Ah, miles has replied. excellent. | 13:53.43 |
kens2 | ah cool | 13:53.58 |
| excellent | 13:54.14 |
henrys | we must have a jbig2 spec about somewhere. Anybody know where it is? (http://www.iso.org/iso/catalogue_detail.htm?csnumber=22394) | 14:35.27 |
jogux | fredross-perry: btw, I think there were some mupdf things you asked for review of a while back, did someone review them for you? | 14:36.06 |
fredross-perry | no not yet. | 14:36.31 |
Robin_Watts | henrys: i have a copy. | 14:36.48 |
kens2 | I have too I htnk | 14:37.12 |
henrys | can it be put on casper? | 14:37.18 |
Robin_Watts | henrys: Uploading as we speak. | 14:37.30 |
henrys | ta | 14:37.42 |
Robin_Watts | jbig2_14492_2001.pdf in my homedir. | 14:37.54 |
| I suspect I got mine from ken. | 14:38.05 |
| kens: Helen has been looking at flights. I think these are her favoured options: https://www.britishairways.com/travel/fx/public/en_gb?eId=111011×tamp=0817022021&source=continue&source=continue | 14:38.51 |
| OK, that link doesn't work :) | 14:39.12 |
| Out on Tuesday LHR->NAS->FPO, then back on monday FPO->MIA->LHR | 14:39.44 |
| The alternative of going back via Nassau involves hanging around in Nassau airport for 9 hours. | 14:40.17 |
| fredross-perry: I've got the latest libgs.so built. | 14:42.00 |
| http://ghostscript.com/~robin/libgs.so | 14:42.13 |
tor8 | Robin_Watts: don't you get the immigration hell in miami? when I check flights I see only a 2h stop in Nassau | 14:45.25 |
Robin_Watts | 3hr stop. | 14:46.09 |
| And if they have the same kiosk system there as they have elsewhere, we should be fine. | 14:46.26 |
tor8 | on sunday, there's an 18:45 bahamasair 326 flight to nassau, then at 21:45 BA252 to london | 14:47.12 |
Robin_Watts | And Professor Google says they do have those kiosks there. | 14:47.16 |
tor8 | kiosk system? | 14:47.39 |
Robin_Watts | tor8: Flying sunday would mean flying out on monday, and no FPO option there. | 14:47.43 |
| tor8: If you've flown into the US using an ESTA on your current passport, then you can use an automated kiosk to get in. | 14:48.06 |
tor8 | I mean flying home on sunday | 14:48.13 |
Robin_Watts | yeah, flying home on sunday would mean flying out there on monday in order to get the 6 nights. | 14:48.54 |
tor8 | Robin_Watts: is that a new thing? I don't think I've seen that anywhere; only the pay-an-obscene-amount-of-cash to get on the trusted traveller or whatever the war on terror term is program | 14:49.11 |
Robin_Watts | Yes, new thing. | 14:49.35 |
| You missed the last staff meeting, hence you've not seen them. | 14:49.49 |
tor8 | just as I discovered the carry-on only line in chicago | 14:50.21 |
| right, staying exactly 6 nights is awkward to fit the nassau flights | 14:51.03 |
jogux | fredross-perry: can you remind us what was ready for review? | 14:58.56 |
fredross-perry | sure, itâs the last four commits in my repos/mupdf.git | 15:01.12 |
| except not the last one, just the three before that. | 15:01.29 |
kens2 | Robin_Watts : we thought we would drop bags at Nassau airport then head into town for the afternoon | 15:04.30 |
| (Sorry Miranda isn't amkign soudns today, I'm not sure why, so missed the ping) | 15:04.52 |
Robin_Watts | kens2: Fair enough, as long as you're aware. Flying via Miami gets you home in a shorter time overall, so we will probably do that. | 15:05.24 |
kens2 | Not a problem. | 15:05.32 |
| We also are lookign at staying an extra night as an option | 15:05.42 |
Robin_Watts | helen (as usual) is tightly constrained on dates. So the tuesday -> monday seems like the best fit for us. | 15:06.35 |
kens2 | No big deal, we don't have ot fly on the same dates after al;l, or evne the same route! :-) | 15:07.07 |
jogux | henrys: does Pete have an Artifex Office365 account? | 15:23.39 |
| Robin_Watts: Two of fredross-pery's commits on http://git.ghostscript.com/?p=user/fred/mupdf.git;a=summary are Android, do you want to review those? | 16:39.42 |
Robin_Watts | if you want. | 16:39.53 |
| The gradle one looks plausible enough. | 16:40.53 |
| The next one falls down on not using tabs. | 16:41.06 |
| but other than the whitespace issues, it looks OK. | 16:43.38 |
jogux | fredross-perry: so the iCloud one; just to follow iOS noms, iCloudEnabled should be BOOL (not Boolean) and should be prefixed _ to help people grok it's a member variable rather than a local | 16:52.25 |
| Robin_Watts: thanks. that helps. my Android is rusty these days :) | 16:52.37 |
| is mupdf K&R style {}s? | 16:52.57 |
Robin_Watts | It's not gnu. | 16:53.11 |
| Generally { on separate lines. | 16:53.26 |
jogux | and no trailing spaces? (is this written down somewhere?) | 16:55.46 |
Robin_Watts | It is not formally written down, but there are commit checks. | 17:01.48 |
jogux | fredross-perry: there's some review comments in the log (I'm still going, xcode is playing up for me...) | 17:14.39 |
| fredross-perry: the iCloud support doesn't seem to work for me, in the app I get 'Cannot open document' and the console says:open document '/private/var/mobile/Library/Mobile Documents/com~apple~CloudDocs/office encryption/[MS-OFFCRYPTO].pdf' | 17:19.59 |
| error: cannot open /private/var/mobile/Library/Mobile Documents/com~apple~CloudDocs/office encryption/[MS-OFFCRYPTO].pdf | 17:20.00 |
| error: cannot load document '/private/var/mobile/Library/Mobile Documents/com~apple~CloudDocs/office encryption/[MS-OFFCRYPTO].pdf' | 17:20.01 |
| 2015-08-17 18:19:22.985 MuPDF[1374:601749] plugin com.apple.UIKit.fileprovider.default invalidated | 17:20.01 |
fredross-perry | feh. | 17:20.23 |
| are you doing this on a device, or in the emulator? | 17:20.50 |
jogux | device | 17:21.00 |
| do I need to mess with entitlements or something first? | 17:21.07 |
fredross-perry | Iâll have to try that then. Doe yu get the iCloud navigator, and used it to pick your document? | 17:21.29 |
jogux | yep, could pick a document fine, but two diferent documents I picked said that (both PDFs) | 17:21.58 |
fredross-perry | ok | 17:22.04 |
jogux | iOS 8.4 on an iPad if that matters. | 17:23.27 |
fredross-perry | OK I am getting this too. | 17:26.33 |
jogux | I think you maybe need to keep the URL around longer? that has the magic token in it that lets us access the file even though it's outside our sandbox | 17:38.50 |
| fredross-perry: anyway, you probably should get distracted from gsproof stuff, I imagine that is more important with a deadline less than a month away :) | 17:49.12 |
| probably shouldn't, even ;-) | 17:49.19 |
fredross-perry | Si. | 17:49.48 |
Robin_Watts | I'm failing to get the gproof stuff working on android locally. | 17:49.55 |
| I've got a libgs.so rebuilt, but it's baling out with a -100 error code. | 17:50.23 |
fredross-perry | what specifically is bailing? | 17:52.03 |
Robin_Watts | The gs invocation. | 17:53.57 |
| It's running gs, but getting a -100 code back. | 17:54.06 |
| What I see on screen is that when I hit the 'proof' button, I get a momentary placeholder page, then that disappears, and I'm left with just the backdrop showing. | 17:54.46 |
| It's as if I've opened a 0 page document. | 17:54.53 |
mvrhel_laptop | Robin_Watts: I am still having issues with the gs api calls in mupdf | 17:55.05 |
Robin_Watts | It's probably a fault in the way I've built the libgs.so | 17:55.11 |
mvrhel_laptop | the init works and then I am getting an io error back from gs when I actually do the arg run | 17:55.30 |
Robin_Watts | An io error ? | 17:55.38 |
mvrhel_laptop | this is with me even hard coding the args to run tiger | 17:55.39 |
| yes error -12 | 17:55.43 |
| you would think I have never run gs before.... | 17:55.57 |
| I am having issues digging in to see exactly what the issue is | 17:56.38 |
| when I said you I did not mean you personally | 17:56.51 |
Robin_Watts | :) | 17:56.59 |
mvrhel_laptop | just frustrated that I can't figure this out.... | 17:57.00 |
Robin_Watts | Can you breakset within gs? | 17:57.19 |
mvrhel_laptop | yes | 17:57.24 |
| listen to this though | 17:57.39 |
Robin_Watts | Oh! I know what I've cocked up. | 17:57.49 |
mvrhel_laptop | I did hook up the stdio with some functions to monitor | 17:57.50 |
Robin_Watts | I'm using file, not memory for clist. Thanks mvrhel. | 17:58.01 |
mvrhel_laptop | oh would this explain my issue? | 17:58.42 |
Robin_Watts | No. Just mine :) | 17:58.54 |
| sorry. | 17:59.11 |
mvrhel_laptop | when I hook up the monitor functions , gs reports back the copyright stuff but then never finishes | 17:59.14 |
| it just goes off and runs and runs and runs | 17:59.22 |
| that is even doing tiger at 96dpi | 17:59.32 |
| no i/o error | 17:59.39 |
Robin_Watts | mvrhel_laptop: That smells like it's waiting for input maybe? | 17:59.43 |
mvrhel_laptop | but never returning either... | 17:59.44 |
| yes | 17:59.47 |
| that was my thought | 17:59.50 |
Robin_Watts | Did you hook up stdin ? | 17:59.52 |
mvrhel_laptop | yes | 17:59.54 |
| but it is never called | 17:59.58 |
Robin_Watts | Did you hook up stdin correctly? :) | 18:00.02 |
mvrhel_laptop | I had a break point there | 18:00.03 |
| ha | 18:00.05 |
| well I think so | 18:00.10 |
| I will keep digging | 18:00.42 |
Robin_Watts | Stupid question (cos this is the kinda think I'd get wrong) did you put the stdin/stdout/stderr handlers in the correct order? | 18:00.56 |
mvrhel_laptop | yes in, out, err | 18:01.47 |
Robin_Watts | mvrhel_laptop: Try running with -Z! ? | 18:02.54 |
mvrhel_laptop | yes let me do that | 18:03.05 |
Robin_Watts | or whatever the one that spews debug out as it runs is. | 18:03.10 |
mvrhel_laptop | Robin_Watts: one question. is instance supposed to be ** or *? | 18:04.36 |
| I thought it should be void *instance | 18:04.49 |
Robin_Watts | In which call? | 18:05.02 |
mvrhel_laptop | in its definition | 18:05.14 |
| you call init it &instance | 18:05.26 |
| s/it/with | 18:05.32 |
| and then you use instance for the others | 18:05.38 |
| this is in generate page in gprf-doc.c | 18:06.16 |
Robin_Watts | You are right, it should be a void *. | 18:06.35 |
| But that should make no difference. | 18:06.39 |
mvrhel_laptop | true | 18:06.44 |
| but it seemed odd to me | 18:06.48 |
Robin_Watts | But I will fix it here, thanks. | 18:06.57 |
| Ok, fredross-perry. I have it "working" under android now. | 18:19.39 |
fredross-perry | meaning...? | 18:19.53 |
Robin_Watts | When I hit the proof button, it goes away and thinks for a bit. | 18:20.08 |
| I then get a completely black page (interestingly, the first time I did it I got a white page) | 18:20.26 |
fredross-perry | hmmm. | 18:20.36 |
Robin_Watts | If I then click the the strange red icon in the top bar, I get an image. | 18:20.57 |
| (with skewed colors, but...) | 18:21.05 |
| If I click it again, I get a different skewed color image. | 18:21.29 |
| and repeated clicks get me to an all white image. | 18:22.18 |
| I'll debug for a bit to see what's going on. | 18:22.33 |
fredross-perry | right. The code there is just trying to cycle thru available separations. I plan to have separate (ha!) buttons. | 18:24.15 |
Robin_Watts | fredross-perry: One problem that occurred to me the other day... | 18:24.40 |
| The 'number of separations' figure is not correct until after the initial render has been done. | 18:25.02 |
| I have to go wire a cooker in. back in a mo. | 18:25.42 |
fredross-perry | yeah, I thought about that too. Thatâs OK, we just need to get that value back out to the activity somehow. | 18:26.03 |
| Is that a euphemism? | 18:26.20 |
mvrhel_laptop | jesus | 18:29.15 |
| oops sorry if I offended anyone | 18:29.31 |
| I just can't believe what I just did | 18:29.39 |
| stupid visual studio was sitting at a break point in the gs thread and never popped up to let me know that it was sitting idle | 18:30.12 |
| that is why it was never returning :0 | 18:30.26 |
| anyway, moving forward now.... | 18:30.51 |
fredross-perry | Robin_Watts: what I have now appears to be working with your latest prebuilt .so file. !!! | 18:35.31 |
Robin_Watts | fredross-perry: Ah, fab. | 18:38.44 |
| mvrhel_laptop: Ah, well, that's good news, I guess. | 18:39.00 |
mvrhel_laptop | any movement forward is good news this morning | 18:39.21 |
Robin_Watts | tor8: INT32_MAX is not defined on windows. | 18:39.24 |
| so current mupdf is broken on windows. | 18:39.32 |
mvrhel_laptop | :( | 18:40.06 |
| Robin_Watts: so do you have fixes in golden for gprf-doc.c that I need to pull | 18:40.28 |
| or are they on your repos? | 18:40.35 |
Robin_Watts | Everything (except the void **instance) is pushed already. | 18:40.48 |
| BUT... don't pull now. | 18:40.54 |
mvrhel_laptop | ok | 18:41.06 |
fredross-perry | so the first image is black (no separation selected). Then as I cycle thru them it seems like I am getting four different âcolorsâ. Not sure what they are. | 18:41.07 |
Robin_Watts | or pull, but be prepared to patch to get it building. | 18:41.14 |
mvrhel_laptop | no I will wait until its all good | 18:41.24 |
| I have plenty to keep me busy | 18:41.29 |
| Robin_Watts: you may want to add -I%romesource/Init/ in the command line | 18:42.49 |
| to ensure that the romfs is always used | 18:42.56 |
fredross-perry | mvrhel_laptop: if youâre working on proofing for gsview, we should talk so I can get that working in the same way, UI-wise. | 18:42.57 |
Robin_Watts | mvrhel_laptop: Can you stick that in an email please? irc eats % quotes. | 18:43.26 |
mvrhel_laptop | oh yes | 18:43.32 |
Robin_Watts | added a unicode version of the getopt stuff. | 18:44.22 |
| for windows. Someone has removed it, and left windows broken. | 18:45.06 |
mvrhel_laptop | fredross-perry: yes, lets plan to talk about the ui | 18:45.59 |
fredross-perry | letâs plan to plan to talk about it. | 18:46.50 |
Robin_Watts | If fact, this code is just utterly broken now. I'm just going to revert the removal. | 18:47.43 |
mvrhel_laptop | lunch time | 19:08.52 |
Robin_Watts | mvrhel_laptop, fredross-perry: OK,4 commits on robin/master | 19:54.02 |
| With those in, I get decent colors out of windows mupdf when coming from separations. | 19:54.31 |
| fredross-perry: In amongst josephs comments on your commits, did you see mine? | 20:03.18 |
| namely that the whitespace convention for the android changes is wrong. | 20:03.33 |
fredross-perry | SO close. I have four buttons hard-wired to the four separations. Which are, Cyan, Magenta, Yellow and Black. If I turn them ALL on, I get just a solid grey page. But if I turn any three on, I get something reasonable-looking. If I turn black off and the other three on, I get what I think it should look like, had I turned black on. | 20:04.29 |
Robin_Watts | Ok, with my latest commits, the android thing seems to work as I expect. | 20:04.41 |
fredross-perry | sorry, not yet. | 20:04.43 |
Robin_Watts | (if each push on the button disables each separation in turn) | 20:04.54 |
fredross-perry | I am assuming that I can enable/disable seps indepently. So my four buttons do that. | 20:05.28 |
Robin_Watts | removing C doesn't make much difference. Removing C+M makes a lot, Removing C+M+Y leaves just greyscale, and then removing C+M+Y+K leaves a blank page. | 20:05.29 |
| fredross-perry: I would grab the latest commits from my repo. Hopefully that might solve it. | 20:06.01 |
fredross-perry | I did that. They seem to be about an hour old? | 20:06.16 |
| when I did that it got much better. | 20:06.31 |
Robin_Watts | Yeah. | 20:06.38 |
fredross-perry | If I remove C+M+Y+K I do get a blank page. And as I turn on C, M, and Y in that order, I get what Iâm expecting. WHen I finally turn K on, I get a grey screen. | 20:07.30 |
Robin_Watts | If you turn them all on, then you should get the rgb data. I have it nobbled here to never use the rgb data. Let me undo that nobbling and retest. | 20:07.30 |
fredross-perry | ahhh. OK. | 20:07.46 |
| is that a eupemism? | 20:07.56 |
| I updated android.zip in my public html area, if you want to grab that to see what I am doing with the four buttons. | 20:11.12 |
| out for a bit, back in an hour or less. | 20:11.24 |
Robin_Watts | ok, the rgb case is indeed knackered. | 20:11.33 |
| I'll look into why tomorrow. | 20:11.37 |
| For now, just disable at least one of the separations and you'll be fine :) | 20:11.54 |
tor8 | Robin_Watts: the unicode getopt was redundant. we have windows specific versions to convert UCS2 to UTF8 which we should use instead. | 21:07.18 |
| though it appears that my patch doesn't use the actual functions we have... fz_argv_from_wargv | 21:11.03 |
| and just reimplements it inline | 21:11.12 |
| I guess I should look into setting up my windows build system so I can fix that properly | 21:12.45 |
mvrhel_laptop | hmm I can't get to the repository for some reason | 21:54.58 |
| oh there it goes | 21:55.23 |
| that was odd | 21:55.25 |
| so is the mupdf golden still broken for windows or are we all good? | 21:57.09 |
Robin_Watts | mvrhel_laptop: If you review my patches, then I can push them, and it will be fine. | 22:20.53 |
| tor8: It also fails to copy the filename and the css filename correctly as far as I can see. It didn't even build :( | 22:21.35 |
| I am quite prepared to believe that with massaging it might work, and then we can ditch the unicode fz_getopt stuff. | 22:22.07 |
fredross-perry | Robin_Watts: OK, one more android.zip for you. I took out the four buttons, and instead have one. When you tap it, you get a popup menu with a list of the actual names of the separations, and check boxes. Iâd like to add a color spot next to each one, as soon as I can trust the rgb color values. Also a helpful header of some sort on the menu would be nice. Finally, using android.widget.PopupMenu requires API 11 as a | 22:22.58 |
| miniumu (weâre at 8 now). Not sure this is OK with everyone. | 22:22.58 |
Robin_Watts | I think there is a vanishingly small number of people using Android < 3 anymore. Certainly the proofing stuff would happen on higher androids. | 22:32.32 |
| If the android build system wasn't so crap, we'd have it use a higher API if we were SUPPORT_GPROOF. | 22:33.14 |
| I'll look into the problem with the RGB data tomorrow. | 22:43.38 |
mvrhel_laptop | Robin_Watts: patches look good | 23:42.36 |
| Forward 1 day (to 2015/08/18)>>> | |