| <<<Back 1 day (to 2015/07/22) | 20150723 |
fredross-perry | How do I make/get a file thatâs suitable for testing gsproof? | 00:47.44 |
wilr | chrisl: Hi, today around noon UTC, I plan to announce the CVE related to bugs 696041 & 696070. Could you please make them public on your bugzilla around that time ? Red Hat will do the same with their bug entry. Thanks | 06:40.26 |
chrisl | wilr: sure, will do | 06:47.12 |
wilr | thank you | 06:49.35 |
Robin_Watts | fredross-perry: Not sure what you mean. Just load any PDF file, and click the 'P' icon. That will save a '.gproof' file into the same dir as the PDF. | 09:30.37 |
wilr | chrisl: could you please allow public access to both bug entries ? | 12:25.21 |
chrisl | wilr: they should be visible now | 12:27.46 |
wilr | thanks ! | 12:28.36 |
Robin_Watts | Morning fredross-perry | 15:14.37 |
fredross-perry | hello. | 15:14.46 |
Robin_Watts | You asked a question last night that I don't really understand. | 15:14.50 |
| You wanted to know how you created a file to test gproof. | 15:15.06 |
| At this stage, any PDF file will do. | 15:15.52 |
| Load it into android mupdf, and hit the 'P' icon. | 15:16.13 |
| That should reload it in proofing mode. | 15:16.19 |
fredross-perry | Does the code create its own separations? Id did hit âPâ. Lts of time goes by andthen the same file seems to appear. | 15:16.40 |
rayjj | Robin_Watts: that just tests mupdf generating the "request" file, right. Then gs has to be invoked with the parameters to generate the actual proof images, right ? | 15:16.57 |
Robin_Watts | rayjj: No. | 15:17.12 |
| When you hit P, mupdf outputs a .gproof file. | 15:17.40 |
rayjj | Robin_Watts: mupdf can't generate the tiled 'proof' images, can it ? | 15:17.49 |
Robin_Watts | The .gproof file contains the path of the PDF, and a list of all the tiled sizes. | 15:17.57 |
| It then invokes itself on that .gproof file. | 15:18.09 |
rayjj | but the .gprf file comes from gs doesn't it ? | 15:18.36 |
Robin_Watts | So a new version of Android MuPDF starts up, and reads the .gproof file. When it tries to display the first page, it calls libgs.so through gsapi, and gets that to render a .gprf file that it then reads in and displays. | 15:18.55 |
| fredross-perry: So what you see as "the same file" is actually "the proofed version of the same file". | 15:19.21 |
rayjj | Robin_Watts: so that's if you have the mupdf built and the libgs.so available | 15:19.31 |
Robin_Watts | rayjj: Right. | 15:19.39 |
fredross-perry | RIght. I can see all of that happening. But what should the user be able to do with this proofed version? | 15:20.07 |
rayjj | the latest Android app that is capable of invoking libgs that is | 15:20.09 |
| fredross-perry: admire it ? | 15:20.25 |
Robin_Watts | rayjj: If you build with SUPPORT_GPROOF it adds the required hooks and includes the libgs.a in the .apk file. | 15:20.53 |
| fredross-perry: When MuPDF opens on a .gproof file, the android UI, should: | 15:21.23 |
| 1) NOT display the 'P' icon. | 15:21.28 |
fredross-perry | I have the right build, using SUPPORT_GPROOF, lib gets loded, etc. | 15:21.29 |
rayjj | if you start with a file that has CMYK colors in it, it should look different to the non-proofed version. IIRC, making a tiger.pdf from tiger.eps will have CMYK colors | 15:21.33 |
Robin_Watts | 2) NOT display the annotation stuff. | 15:21.51 |
| 3) NOT display search or reflow etc. | 15:22.10 |
| 4) display a list of the separations from the each page, and enable them to turned on an off. | 15:22.38 |
rayjj | fredross-perry: that's because gs will generate the tiger.pdf using the iccprofiles/default_cmyk.icc | 15:22.44 |
| Robin_Watts: the list of separations includes the CMYK colorants ? (but not RGB) | 15:23.51 |
fredross-perry | OK I think I get it now. Is there gs API for rendering indivdual separations yet? | 15:24.14 |
Robin_Watts | fredross-perry: .gprf files contain both an RGB version of the image, plus a load of individual separations (typically, but not always CMYK, or CMYK + some extras) | 15:24.27 |
| At the moment the rgb version is generated very craply from the CMYK versions. | 15:24.52 |
rayjj | Robin_Watts: when won't it have CMYK ? | 15:24.58 |
Robin_Watts | But michael is going to fix that so it's all color correct. | 15:25.03 |
| rayjj: I believe we can have cases (or may be able to have cases) where files don't use CMYK, but rather use duotones etc. | 15:25.35 |
| but I won't swear to that for now. | 15:25.46 |
rayjj | Robin_Watts: oh, so since it uses the crap CMYK->RGB, what I said about tiger.pdf looking different is not yet true | 15:25.47 |
Robin_Watts | rayjj: The crap CMYK -> RGB conversion done in gs when creating the .gprf file does not give the same results as the direct rgb rendering that mupdf does. | 15:26.55 |
| So the tiger does look different currently, but not for a particularly great reason :) | 15:27.14 |
rayjj | Robin_Watts: since DeviceN devices are always subtractive CMYK + optional separations, the only way it wouldn't have CMYK is if the gs device detected that a color was unused and omitted it. | 15:27.16 |
Robin_Watts | rayjj: Right. My point is that the .gproof format is designed not to treat CMYK specially. | 15:27.46 |
| Hence as far as fred is concerned, CMYK are just four of the possible separations he's getting in. | 15:28.08 |
rayjj | Robin_Watts: is the crap CMYK->RGB used currently for gprf different than the crap CMYK->RGB that mupdf uses ordinarily ? (neither use ICC profiles, AFAIK) | 15:28.28 |
Robin_Watts | rayjj: can't remember. It's a moot point as Michael is doing a better version. It's just a stopgap. | 15:29.44 |
| So, fredross-perry.... Yes, there is an API for changing the separations displayed, but it's at the MuPDF level, not the gs one. | 15:30.31 |
| gs always renders all the separations to the file. | 15:30.37 |
| MuPDF picks up just the currently selected ones from the file. | 15:30.55 |
fredross-perry | OK, this gives me enough to go on I think. Thanks all | 15:30.59 |
Robin_Watts | (if all separations are selected, it uses the RGB values) | 15:31.10 |
| (if not all the separations are selected, it picks up the individual separations and combines them together itself to form it's own (non-color correct) RGB version) | 15:31.47 |
rayjj | Robin_Watts: is that subset combo blending implemented yet ? | 15:32.23 |
Robin_Watts | There should be MuPDFCore functions you can call to ask for details of the separations on the current page. | 15:32.30 |
| There are calls for: How many separations? Enable/Disable Separation n. Get me the name/equivalent rgba/cmyk values for separation i. | 15:33.12 |
| When you enable/disable a separation, the core of MuPDF invalidates all the cached fz_bitmaps for that page. | 15:34.37 |
| This means the next time a page is rendered, it will pull the requested separations from the file, and your disabling/enabling will take place. | 15:34.58 |
| Note, that the fz_bitmaps being invalidated does NOT mean that the java level bitmaps will have been invalidated. | 15:35.21 |
| Hence when you enable/disable a separation, you'll need to force the java side to invalidate the cached bitmaps it has for the page too. | 15:35.52 |
| rayjj: I have coded the blending of enabled separations to give RGB, yes. | 15:36.34 |
rayjj | Robin_Watts: cool! | 15:36.54 |
Robin_Watts | it is marvellously untested at the moment, as a) I have no UI to cause it to be called, and b) all the equivalent cmyk colors in the gproof device are coming out as 0. | 15:37.16 |
henrys | wow french fans are a tough crowd, throwing urine at chris fromme. don't like the brits winning their race... | 15:40.31 |
chrisl | He probably prefers that to being accused of doping....... | 15:41.10 |
henrys | you know the steroids for the asthma stuff ... shoujld that be okay? A future of asthmatic tour leaders. | 15:42.23 |
rayjj | henrys: well, it does open up the passages, so in theory, it allows artificially induced improved breathing | 15:43.39 |
| henrys: but as long as they allow anyone to use asthma meds, then it's fair, right ;-) | 15:44.44 |
henrys | rayjj: they have to have been diagnosed with asthma | 15:45.08 |
rayjj | henrys: I'm sure that can be arranged with the correct choice of MD ;-) | 15:46.24 |
| after all, asthma can develop at any age due to histamine response | 15:47.39 |
henrys | you know I've watched the tour for years and just loved it but I just can't muster much enthusiasm any more. | 15:47.41 |
chrisl | Huh, just 'cause the Brits are doing well now? ;-) | 15:48.10 |
fredross-perry | Would that âur de franceâ? | 15:49.10 |
henrys | now they're are catching guys with tiny motors that put out an extra 10 or 20 watts. It's just ridulous. My friend works with Greg Lemonde and he's going over there with iron filings to detect the magnets. | 15:49.35 |
| chrisl: no I like most of the brit cyclists | 15:50.22 |
Robin_Watts | henrys: a) What sort of battery do they use? and b) What sort of competitor is going to let anyone near their bike with iron filings? | 15:50.41 |
| fredross-perry: haha. missed that. very good. | 15:52.10 |
henrys | Robin_Watts: anyone the bikes are next to people throughout the race | 15:52.19 |
chrisl | henrys: TBH, I don't enjoy any road racing these days: the speeds are so high (especially on the mounting stages), they're running in such close proximity and there seems to me to be more contact these these days, I find it a miracle there haven't been more serious injuries. And that makes me cringe. | 15:52.33 |
fredross-perry | On another topic, I just got a new macbook, and discovered that I need to make so changes to GSView for Retina. Not hard to do. But, I wonder about and inux boxes with Retina-like screens. Do we have one? | 15:52.50 |
Robin_Watts | fredross-perry: I have a retina-like screen on my laptop, and I might have a linux VM on there. | 15:53.44 |
| Certainly I could put one on. | 15:53.51 |
henrys | chrisl: today descent by Bardet was truly frightening and amaziing - he used every inch of the road. amazing bike handling | 15:54.01 |
Robin_Watts | (actually, my screen is not *that* hi-res, just 1920x1080 in a 13" diagonal, but it's enough for windows to do the rescale stuff) | 15:54.38 |
fredross-perry | I can do that too. But I donât know if Iâll get native screen resolution inside the VM or not. Opinion? | 15:54.38 |
Robin_Watts | fredross-perry: I would have hoped you should. What VM system do you use? | 15:55.10 |
fredross-perry | Virtualbox | 15:56.14 |
chrisl | henrys: there was a stage in one of the Spanish events last year, there was descent and after the leaders went through, one of the spectators dumped a load of water across a corner - luckily, it was spotted by a team car, and they relayed the information back, but it could have been *really* bad | 15:56.19 |
Robin_Watts | fredross-perry: https://www.virtualbox.org/ticket/10860 | 15:57.00 |
| sounds like VMware gets it right, and virtualbox doesn't. | 15:57.16 |
fredross-perry | hmm. Iâll ave to switch then | 15:57.34 |
| thanks | 15:57.36 |
paulgardiner | I've never understood why someone would want their team to win unfairly. Where's the achievement? | 15:57.40 |
Robin_Watts | sponsorship. prizes. | 15:58.26 |
chrisl | Robin_Watts: possibly for an insider, but for a fan? | 15:58.55 |
henrys | Robin_Watts: here's a simple setup http://cyclingtips.com.au/2015/04/hidden-motors-for-road-bikes-exist-heres-how-they-work/ but LeMonde told my friend there are better systems sit in the rim of the wheel are are undetectable except they have a magnet. | 15:59.04 |
Robin_Watts | chrisl: I meant insiders. | 15:59.12 |
henrys | chrisl: I know this year they had a bunch of cows wander out on the descent, they got around them but ... | 16:00.23 |
| that said Lemond is obsessive about catching cheaters in bike racing he told people for years Armstrong was cheating and he lost everything doing it... | 16:02.12 |
chrisl | henrys: I really wouldn't like to try to make evasive manoeuvres at 50+ mph on those skinny, ultra-high pressure tyres! | 16:03.12 |
| Possibly Robin_Watts was picturing something more like this: http://www.southbayriders.com/forums/attachments/16355/ | 16:05.32 |
henrys | right in a bike race 5 or 10 watts is huge... these guys are really closely matched. | 16:06.24 |
Robin_Watts | chrisl: No, I was just pondering whether the benefits of the motor drastically outweigh the weight of the battery. | 16:06.33 |
| I can't believe a battery would power you for a large percentage of the race, and then you've got to drag the battery around for the rest of the time. | 16:07.15 |
chrisl | Robin_Watts: I felt the same thing about KERS, but...... | 16:07.45 |
Robin_Watts | chrisl: Well, in car racing you spend lots of time on the brakes... | 16:08.11 |
| the kers is used/recharged (mostly) every lap. | 16:08.54 |
chrisl | Robin_Watts: I still wondered about the substantial extra weight of the entire system being worth it - clearly, I was proven incorrect! | 16:09.15 |
Robin_Watts | on a bike, are there really periods of time when the rider has 'spare' capacity to recharge the battery? | 16:09.22 |
chrisl | It would be hard to have an energy scavenging system without it being visible | 16:10.27 |
| With this kind of thing, there should be a total no-tolerance - if you're caught with that kind of system on you bike, the entire team gets a lengthy ban | 16:13.55 |
henrys | anyway it's all the rage if you google for "mechanical doping" ... you're probably overestimating weight bikes have to be a minimum weght to race and manufacturers are well below those standards so the bike can have the battery without weight cost | 16:15.05 |
Robin_Watts | henrys: ah, yes, of course. | 16:15.22 |
henrys | lots of people citing this at the vuelta ... hard to tell it does look suspicious: http://www.cyclingweekly.co.uk/news/latest-news/fresh-motorised-doping-claims-ryder-hesjedals-bike-moves-134745 | 16:30.51 |
| chrisl: did you see mvrhel_laptop's notes about jasper - we do want that removed right? | 16:35.44 |
chrisl | henrys: yes, he did it | 16:36.08 |
Robin_Watts | Did he also remove lcms v1 ? | 16:36.19 |
chrisl | Yes | 16:36.26 |
Robin_Watts | But we still ship that, right? | 16:36.34 |
chrisl | Nope | 16:36.41 |
Robin_Watts | oh, ok, didn't realise that had gone. | 16:36.52 |
chrisl | I think, in theory, you could still link to lcms1 but we haven't shipped for quite some time | 16:37.36 |
mvrhel_laptop | chrisl: I guess we should go ahead and remove the file gsicc_lcms.c which was the interface to lcms1. I don't see any reason to keep it | 17:19.22 |
| you all may enjoy reading this. http://tinyurl.com/ob5fwdt | 17:41.52 |
| Hmmm a little surprised with IEEE and (Access to the Web app is US $0.99.) Considering the amount I pay for dues each year.... | 17:43.51 |
Robin_Watts | mvrhel_laptop: C is a huge improvement on all it's successors. | 17:46.45 |
| fredross-perry, mvrhel_laptop: Back to Android gproof for a mo, if I may. | 17:50.50 |
| The MuPDFCore class keeps 3 fz_pages on the go at any one time. The current one, the one before, and the one after it. | 17:51.34 |
fredross-perry | in a meeting, get back to you in 20 or so? | 17:51.35 |
Robin_Watts | fredross-perry: Sure. I'll burble here, just ignore me til you're done. | 17:51.48 |
| The selection of which separations to enable is done at the fz_page level. | 17:52.26 |
| This means that if you proof a PDF, then choose the separations you want on page 1, then flip to page 2, then flip back to page 1, you'll still see the same selections of separations that you had before. | 17:53.29 |
| If you flip to page 3, then back to page 1, page 1 will have been discarded and reloaded, so the separation selection will have been lost. | 17:54.14 |
| I suspect we need to make the viewer remember the separation selection for each page independently of the fz_page. | 17:54.56 |
| Then whenever we reload a page, reset the selection before we run that page. | 17:55.13 |
| When I started burbling I was going to suggest that that should be done at the java level, but I think I've now convinced myself it should be done in the JNI layer (mupdf.c) | 17:56.19 |
| fred can probably ignore the problem for now, and one of us can fix it later - it shouldn't change the java API. | 17:57.04 |
| But I wanted to flag it up while I remembered it. | 17:57.17 |
fredross-perry | rhanks, I think I understand that. | 18:03.33 |
Robin_Watts | Henrys: https://video-lhr3-1.xx.fbcdn.net/hvideo-xpf1/v/l/t42.1790-2/1464125_566708843398824_809937593_n.mp4?efg=eyJybHIiOjcxOCwicmxhIjo4NTR9&rl=718&vabr=399&oh=26b246e487532f5812f79349cbc288a8&oe=55B15515 | 18:42.22 |
| Get Sabrina one, and she can race you. | 18:42.57 |
henrys | Robin_Watts: I wonder if inexperienced riders need more power | 20:07.58 |
| fredross-perry: you aren't by chance in the meeting with miles? he just called me and asked me to call him back but his phone just rings once ? | 20:09.11 |
fredross-perry | no I am not, sorry. | 20:19.39 |
mvrhel_laptop | Robin_Watts: I accidently pushed my gproof update to the main repos. I meant to push them to my repos on casper for you to look over. they are pretty straight forward. let me know if you have an issue | 20:40.31 |
| hmm somehow my remote repos is screwed up | 20:42.39 |
| need to eat lunch now | 20:42.51 |
| ok. nothing that a forced push did not fix... | 21:24.02 |
| now the rest of the changes I will stage for you to review Robin_Watts | 21:24.58 |
Robin_Watts | mvrhel_laptop: cool. | 23:14.00 |
| The stuff you've pushed looks good (to my limited understanding). Looks like a lot less complexity in the device. | 23:14.23 |
rayjj | isn't complexity good for job security ;-) | 23:25.22 |
rayjj | channels former employees that had that opinion | 23:25.51 |
| NB, none of them worked for me long it they didn't change their attitude. Artifex struggles with that. | 23:26.50 |
| there was one really sharp guy that came in from college and all of his variable names and branch destinations (ASM) were girls names. And no comments. | 23:28.09 |
| After 2 weeks, and looking at his work for the first time, I had to give him a choice. Luckily for me, he chose to "get real" and use useful variable names and branch target names and stayed 11 years and became a 'star' engineer. I guess school didn't teach him about that | 23:30.05 |
| after all, in school, most assignments are "throw away" | 23:30.53 |
| Forward 1 day (to 2015/07/24)>>> | |