| <<<Back 1 day (to 2014/08/25) | 2014/08/26 |
hdon | hi all :) i'm trying to learn postscript using gs interpreter | 02:47.26 |
| is "load" not available in gs? | 02:47.33 |
| i get an error when i try to use "load" like so: << /foo 42 >> /foo load | 02:49.28 |
| ohh... i need to use "begin" to put it on the dictionary stack | 03:13.18 |
Robin_Watts | tor8: Morning. | 08:29.09 |
| Did you draft a reply to the epub stuff yet? | 08:29.26 |
tor8 | Robin_Watts: morning. no. | 08:29.44 |
| give me some time to wake up and I'll get back to you with a draft | 08:30.24 |
Robin_Watts | cool. | 08:31.08 |
Robin_Watts | is in SOT hell, so looking for displacement activities :) | 08:31.20 |
pedro_mac | you love it really | 08:31.38 |
tor8 | Robin_Watts: I've written up some notes on epub and epub3 here: http://ghostscript.com/~tor/epub.txt | 10:09.56 |
Robin_Watts | Why are drop shadows impossible? Just because we don't support them in the engine, currently? | 10:11.48 |
tor8 | Robin_Watts: yeah, and they're part of a family of CSS 3 extensions that are just plain insanity | 10:12.13 |
| text outlines probably would work with the text rendering modes | 10:12.28 |
| I'm just putting down my foot and saying No! here on some of these items :) | 10:12.40 |
| ruby text should be possible, but it's a vast time sink with dubious returns | 10:12.59 |
Robin_Watts | I would be tempted to characterise Audio/Video and speech as being 'possible, but not down to us'. | 10:13.00 |
| i.e. we can hand them off to app supplied plugins. | 10:13.20 |
| but otherwise, looks good. | 10:13.49 |
tor8 | Robin_Watts: yeah. those two are probably the ones we might want to expose in some way. | 10:14.02 |
| because it sort of makes sense to allow embedded video and audio in a fancy text-book | 10:14.24 |
| I don't look forward to writing yet another parser (HTML5) so soon after doing JavaScript :( | 10:15.15 |
jogux | shudders at the thought of writing a css/layout engine. at least it sounds like it's not full css/html5 :) | 10:24.22 |
tor8 | Robin_Watts: hm, it looks like EPUB3 only uses a subset of CSS3 -- that makes things a bit easier | 10:26.42 |
sebras | tor8: hey, the js parser didn't take too much time. | 12:54.14 |
| tor8: but html is probably less strict. :/ | 12:54.30 |
kens | rayjj I think there may be a problem with the addition of OutputICCProfile to the dynamicppkeys dictionary, though I need to do more work on it | 14:06.31 |
rayjj | kens: what kind of problem. As you said, it seems that always getting params from the device should work better. | 14:07.38 |
kens | If I read the code correctly, the keys in the dynamicppkeys dictionary are not fed to the device..... | 14:08.05 |
rayjj | kens: the problem with OutputICCProfile is (I think) that the ColorConversionStrategy distillerparam has the side effect of changing OutputICCProfile | 14:08.23 |
kens | With the exceptionof LeadingEdge, which is no longer an entry in dynamicppkeys anyway..... | 14:08.24 |
| rayjj that's what causes the problem with OutputICCProfile, yes. | 14:08.52 |
rayjj | kens: oh, great. I didn't realize that they were treated as "read only" dynamic elements | 14:09.09 |
kens | But I'm unconvinced that's the latest problem you've uncovered | 14:09.15 |
| rayjj, Neither did I :-( | 14:09.23 |
| I'm still not entirely certain, this is a can of worms :-( | 14:09.35 |
| If you look in gs_setpd.ps at around line 455 you'll see the comments 'Define the parameters which should not be presented to the device' and the entire content of dynamicppkeys is enumerated there | 14:10.31 |
| But that routine is called from .prepareparams | 14:11.07 |
| I *think* that its always called (from .trysetparams) but I could be mistaken | 14:11.36 |
rayjj | kens: well, it should be easy enough to test. Just a sec | 14:12.26 |
kens | This is separate from teh bug you raised though. The rangecheck is caused by the fact that the dictionary we are presenting has RedValues (and Green, Blue) with values which are not 1. THis causes an error, I'm not sure why it doesn't generally do so | 14:12.34 |
| I htink <</OutputICCprofile (blah)>> setpagedevice should be enough. Not sure if -sOutputICCProfile will work the same. | 14:14.00 |
rayjj | kens: yes, it does break: gswin32c -c "<< /OutputICCProfile (iccprofiles/sgray.icc) >> setpagedevice" -f examples/tiger.eps | 14:15.53 |
kens | FWIW if, instead of <</ProcessColorModel /DeviceGray>> setpagedevice I do -dProcessColorModel=/DeviceGray then your file works without error. | 14:15.57 |
rayjj | kens: before the change (I have a "reference" binary laying around) my example comes out gray | 14:16.46 |
kens | Right, so it does get filtered out :-( | 14:16.59 |
henrys | marcosw, jogux : not sure if marcosw can attend the skype meeting he usually has to leave early, but I did want to talk about the test framework. marcosw when is a good time for you. The skype meeting is right after mupdf and gs. | 14:17.18 |
rayjj | kens: so treating it the same way as LeadingEdge should work, right ? | 14:17.34 |
kens | We cna overcome it (I believe) weith the same hack used for /LeadingEdge (which is no longer required, as LeadingEdge is no longer in dynamicppkeys) | 14:17.45 |
| rayjj yeah that's what I was trying to say | 14:17.55 |
rayjj | echo echo echo | 14:17.58 |
kens | When I get to the bottom of the RedValues (Green, Blue..) I'll fix ti at the same time | 14:18.31 |
kens | thinks this dynamicppkeys has all the hallmarks fo a dreadful hack | 14:18.59 |
jogux | henrys , marcosw : fair enough. I'm around for the next few hours anyway, so whenever. | 14:22.09 |
rayjj | kens: so the rangecheck error is related to bug 695416 ? | 14:22.56 |
kens | Umm, I don;t think so that's a MuPDF bug | 14:23.25 |
| Its 'kind of' related to 695420 | 14:23.42 |
henrys | Hi fredross-perry welcome | 14:24.16 |
kens | Similar in that we are using our stored page device dictioanry to set values, and those values are no longer valid (or something like that) | 14:24.18 |
rayjj | kens: oops. I meat bug 695439 (I had the other one open in my browser and type the wrong number) | 14:24.29 |
Robin_Watts | deduces that fredross-perry is the gtk guy? Welcome. | 14:24.50 |
kens | waits for Bugzilla to catch up | 14:24.50 |
henrys | for those of you that don't know fredross-perry is working on porting mvrhel_laptop gsview gui to qt. | 14:25.02 |
Robin_Watts | qt, sorry. | 14:25.14 |
kens | rayjj yeah that one | 14:25.16 |
fredross-perry | Well, Qt, but yes! | 14:25.20 |
mvrhel_laptop | henrys: I chatted with him last wekk | 14:25.21 |
| week | 14:25.23 |
rayjj | is glad chatzilla autocompletes names :-) | 14:25.25 |
fredross-perry | hi Michael. | 14:25.30 |
mvrhel_laptop | hi fredross-perry | 14:25.43 |
rayjj | hi, fredross-perry | 14:25.46 |
fredross-perry | hello everyone. | 14:25.56 |
kens | We thorw an error (*not* from pdfwrite, but from gx_default_...) because we have RedValues, the value is not 1 and we say 'its already set', which causes the code to throw a rangecheck. I've no idea yet what this all means | 14:26.08 |
rayjj | kens: I was seeing the ilocate error very consistently with -Z? | 14:26.55 |
henrys | tor8: are you about yet? | 14:27.16 |
kens | rayjj I can now get it, its a slightly different incantation for me | 14:27.18 |
rayjj | kens: OK. | 14:27.33 |
kens | But not radically so. I have no idea what its poiting at, but I'm suspicous its still something related to colour model/ICC profeils | 14:27.47 |
henrys | 3 minutes to the meeting getting some more coffee | 14:27.51 |
fredross-perry | iâll take a cup please. | 14:28.03 |
kens | Ah, another Mac user | 14:28.10 |
marcosw | henrys: I have time to attend the skype meeting today. | 14:28.11 |
rayjj | does changing the ProcessColorModel also have side effects, such as with OutputICCProfile ? | 14:28.34 |
kens | rayjj It must do, I just haven't worked out what yet | 14:28.49 |
henrys | marcosw: great but I don't want you cutting school. | 14:29.29 |
kens | It doesn't (apart from the ICC profile) in pdfwrite, but I think the fact that we store a copy of the page device dictioanry instead of actually using what the device sends back is a recipe for this kind of disaster | 14:29.30 |
henrys | ;-) | 14:29.38 |
kens | Ah, 2 marcosw's; one can go to school and the other can attend the Skype meeting | 14:30.26 |
henrys | fredross-perry: do you know marcosw ? | 14:30.53 |
fredross-perry | No I do not. Hi Marcosw. Where is everyone? Iâm in N. Cal. | 14:31.19 |
rayjj | my twins are bummed. First day of school tomorrow AM | 14:31.20 |
| that's where Marcos is | 14:31.30 |
henrys | can't believe this summer is over, ugh | 14:31.37 |
kens | Me, chrisl, Robin_Watts, paulgardiner are in England, jogux, pedro_mac are in Scotland. Tor8 is in Sweden | 14:32.03 |
rayjj | fredross-perry: I'm in Orange County (Cypress) -- about 20 min from Disneyland by surface streets | 14:32.10 |
fredross-perry | I used to live in SoCal. We lived at dineyland. | 14:32.44 |
henrys | fredross-perry: I asked about marcosw because he's in Fremont and friends with miles. | 14:32.46 |
fredross-perry | I donât recall meeting. Did we? | 14:33.05 |
Robin_Watts | oh, you'd remember :) | 14:33.15 |
fredross-perry | Maybe it was the night of the Rave. | 14:33.25 |
henrys | rayjj: did we ever get ron set up with domain info? | 14:33.32 |
rayjj | fredross-perry: the annual passes have gotten too pricey -- now we just do Knotts (it's also closer -- 10 minutes max) | 14:33.59 |
| henrys: yes, all done | 14:34.08 |
henrys | great | 14:34.14 |
rayjj | he caught me here on IRC last Friday and reminded me | 14:34.37 |
tor8 | henrys: I'm here, sorry had the IRC window obscured | 14:34.56 |
kens | Hmm, I hadn't heard of Knott's berry farm before | 14:35.00 |
henrys | rayjj: miles is a bit uppity about this board. Do you have any schedule guestimates? | 14:35.18 |
fredross-perry | I never went to KBF in all those years. | 14:35.18 |
Robin_Watts | edits logs. | 14:35.33 |
rayjj | it's much less expensive, and they have a water park adjacent and we can get an annual pass to both for less than a single day at Disneyland | 14:36.05 |
mvrhel_laptop | I went to KBF when I was a kid (like 13 years old) | 14:36.14 |
henrys | tor8: fredross-perry had a question about rar and how to get cbz to work properly. Do we require a rar implementation from the platform? | 14:36.24 |
rayjj | mvrhel_laptop: it's a LOT different than then. MANY more roller coasters | 14:36.42 |
kens | thinks the water park sounds more fun | 14:36.57 |
mvrhel_laptop | ok I recall there being one that did a loop and that was about it | 14:37.00 |
rayjj | mvrhel_laptop: even two for smaller kids in "Camp Snoopy" | 14:37.02 |
tor8 | henrys: RAR is a patent and copyright and licensing encumbered format | 14:37.03 |
fredross-perry | CBZ is working, but it turns out the Mac Finder compression uses a variant of the ZIP format that we donât handle. Iâm looking into that. | 14:37.26 |
tor8 | so we don't support cbr files, only cbz | 14:37.28 |
| fredross-perry: it might be 64-bit zips? in that case, I think it's an easy fix | 14:37.43 |
fredross-perry | no, thereâs a variant where the compressed and uncompressed size of each file in the archive is not stored where we expec. | 14:38.19 |
| t. | 14:38.26 |
tor8 | but if you could link (or mail) a file I can look at | 14:38.27 |
fredross-perry | @tor8 whtâs your email address? | 14:38.46 |
henrys | we've siced the lawyer on the NVAreader folks, FYI. I don't know how it will go in Viet Nam, not sure what the IP laws and enforcement is like there | 14:38.54 |
tor8 | fredross-perry: tor dot andersson at artifex dot com | 14:38.57 |
kens | I suspect the best we cna hope for is to get them pulled from the app stores | 14:39.16 |
fredross-perry | okee.dokee | 14:39.22 |
rayjj | fredross-perry: most of us also have just our first name as a an alias, too | 14:39.34 |
fredross-perry | thanks. | 14:39.52 |
henrys | rayjj: so you're getting the ------ board but ++++++t has priority. | 14:40.35 |
| mvrhel_laptop talked to you I assume. | 14:40.43 |
kens | more log editing ? | 14:40.46 |
mvrhel_laptop | yes. all the app stores should respond to the NVA issue | 14:40.46 |
Robin_Watts | kens: I think so. | 14:40.53 |
mvrhel_laptop | oh henrys | 14:41.01 |
fredross-perry | who know the most about how fonts/font code is built in mPDF? Questions offline. | 14:41.11 |
henrys | mvrhel_laptop: that's what she said ;-) | 14:41.20 |
rayjj | henrys: OK. Does the M board have any documentation ? | 14:41.24 |
| which of us will edit the logs ? | 14:41.51 |
fredross-perry | 11 minutes until the first TWSS. Average? | 14:41.52 |
Robin_Watts | rayjj: I have. | 14:41.59 |
tor8 | fredross-perry: that would probably be me... | 14:42.10 |
rayjj | Robin_Watts: that was quick. Thanks | 14:42.19 |
henrys | fredross-perry: tor8 is the person to talk to, fine to shoot him a question here. Usually if you have a question someone else will learn something too. | 14:42.21 |
fredross-perry | great thanks | 14:42.37 |
mvrhel_laptop | rayjj: I will send you all the info. we will be on their external network which will have everything you need | 14:42.48 |
tor8 | fredross-perry: and if I'm not around (or paying attention) someone else might pipe up as well | 14:42.57 |
fredross-perry | grand | 14:43.07 |
rayjj | mvrhel_laptop: OK, that sounds good. | 14:43.08 |
mvrhel_laptop | rayjj: just need the best address to send it to you | 14:43.10 |
Robin_Watts | fredross-perry: So you know, Tor is the original author of MuPDF. Myself, Paul and Michael have also worked on it, but we're on other stuff at the moment. | 14:43.19 |
henrys | fredross-perry: lots of font people about. | 14:43.26 |
rayjj | the 5142 Evergreen Ave is the best | 14:43.37 |
kens | chrisl and I can both answer font questions, but not so much MuPDF ones | 14:43.41 |
rayjj | 90630 | 14:43.44 |
chrisl | I assumed it was mupdf specific font stuff..... | 14:44.15 |
fredross-perry | OK thanks. Iâve got to formulate a detailed question first. | 14:44.22 |
mvrhel_laptop | rayjj: ok will get it out today fedex | 14:44.34 |
henrys | Robin_Watts, tor8 : do we have an answer for epub? | 14:44.43 |
rayjj | mvrhel_laptop: np. | 14:44.44 |
| thanks | 14:44.49 |
tor8 | henrys: I have some ramblings about epub on http://ghostscript.com/~tor/epub.txt | 14:45.08 |
henrys | Robin_Watts, tor8 you guys were going to send something to the customer right? | 14:45.17 |
Robin_Watts | henrys: I was leaving the reply to tor8. | 14:45.25 |
tor8 | the inclusion of css3 is scary, but it's a subset so we might be okay | 14:45.27 |
henrys | tor8: great I'll read that. | 14:46.04 |
tor8 | I was hoping henry would massage my ramblings into something that doesn't over promise or scare the customer | 14:46.05 |
henrys | tor8: sure. | 14:46.25 |
chrisl | CSS3 includes ruby? WTF?? | 14:46.42 |
tor8 | chrisl: ruby as in annotations-of-a-word in japanese, not as in the programming language | 14:47.09 |
chrisl | Ah, that makes more sense...... | 14:47.21 |
henrys | tor8: customer aside I guess I feel prettey strongly we should have this, tor8 but I don't know if epub is losing ground, 2 years ago it would have been a strong addition to our product line. | 14:47.44 |
tor8 | henrys: I expect epub is gaining ground, it's become the defacto standard for all ebooks. even kindle's new KF8 format is an epub in an amazon wrapper. | 14:48.42 |
fredross-perry | Youâve got Speech under âimpossible featuresâ, but it might not be that scary. What platform? | 14:49.14 |
tor8 | in a few years, if they got their act together, I'd expect all the devices to just embed a web browser and use that to render the books. | 14:49.21 |
chrisl | fredross-perry: cross platform | 14:49.44 |
tor8 | fredross-perry: that's the thing ... we don't do audio. it might be possible to ship it off via callbacks or some plugin interface. | 14:49.54 |
fredross-perry | so mobile (both) and desktop (all three)? | 14:50.06 |
tor8 | and audio is not cross platform, and requires us taking over an event loop, etc, etc | 14:50.27 |
rayjj | tor8: the problem I've had reading HTML books on tablets is that they don't have a "page forward" scroll | 14:50.35 |
tor8 | mobile (all three) and desktop (all three, and the bsds) | 14:50.49 |
fredross-perry | Every platform does audio, so you should assume that partâs doable. There are also cloud-based speech renderers (our product is using one now). | 14:51.05 |
tor8 | rayjj: yeah. there are solutions to that, but they depend on the whole "getting their act together" :) | 14:51.16 |
fredross-perry | Bipolar Spectrum Diagnostic Scale? | 14:51.37 |
tor8 | fredross-perry: anything is doable, we just judge it out of our scope | 14:51.47 |
| consider that this project will get one, maybe two, developers on it | 14:52.36 |
henrys | fredross-perry: does Qt have anything to say about cross platform printing? | 14:52.36 |
Robin_Watts | fredross-perry: At heart MuPDF is a portable C library that does file manipulation/rendering. | 14:53.14 |
| We have various example apps that wrap it into tools and viewers. | 14:53.25 |
fredross-perry | They do have it. I built it and tested on Mac so far. Need to try Linux next. | 14:53.25 |
henrys | rayjj: anything for the C schedule? | 14:53.27 |
chrisl | henrys: perhaps you shouldn't be mentioning these companies by name? | 14:53.56 |
Robin_Watts | as such, we try and keep the viewers as 'thin' as possible, so the smarts all stay in the central lib. | 14:53.57 |
mvrhel_laptop | more editing... | 14:53.58 |
rayjj | henrys: still struggling with it, sorry. T | 14:53.59 |
kens | thinks henrys is trying to give the names away.... | 14:54.02 |
henrys | they aren't a customer do they want to be secret if so we can edit the logs. | 14:54.45 |
mvrhel_laptop | I don't know what the NDA says between the two | 14:55.11 |
chrisl | Well, it would be easier to get shot of using "company C/M" etc....... | 14:55.19 |
Robin_Watts | logs edited. | 14:55.24 |
henrys | will do. | 14:55.33 |
rayjj | henrys: I don't know if Miles did any NDA. The stuff I got was not identified as covered material (as it would have to be even with an NDA) | 14:56.01 |
Robin_Watts | fredross-perry: Every time we decide to do a new feature in MuPDF we try to avoid pulling in more dependencies etc. | 14:56.24 |
henrys | nothing else from me for this meeting except review your important project(s) in the agenda. | 14:56.44 |
Robin_Watts | As such audio/video would need to be done by exposing the data via an API and then having the app do the heavy lifting. | 14:56.51 |
fredross-perry | Understood. | 14:57.09 |
mvrhel_laptop | Robin_Watts: with the text extraction of mupdf, it would not be too hard to hook up a 3rd party voice synth. capability to an app | 14:57.40 |
henrys | rayjj: just focus on the conexant board kick everyting else back to me or if you know who willl work on it, assign it to them. | 14:57.48 |
Robin_Watts | mvrhel_laptop: Right. | 14:57.58 |
rayjj | henrys: ok | 14:58.08 |
| sure would be nicer if it was linux :-( | 14:58.46 |
chrisl | rayjj: so what's the stumbling block on it? | 14:58.49 |
fredross-perry | signing off for now.... | 14:58.57 |
henrys | off to the skype meeting. | 14:59.01 |
chrisl | fredross-perry: bye! | 14:59.04 |
kens | bb fredross-perry | 14:59.04 |
fredross-perry | cheers all | 14:59.18 |
chrisl | Hmm, "Skype can't connect....." - good job I'm not needed in the SOT meeting | 15:00.42 |
mvrhel_laptop | rayjj: you will like the company M board then | 15:01.12 |
rayjj | mvrhel_laptop: I'm sure it will be better | 15:01.26 |
mvrhel_laptop | I know miles really wants to get the C board done though and then shipped of to japan I suspect | 15:02.06 |
| s/of/off/ | 15:02.15 |
| there really is a small window for this | 15:02.36 |
rayjj | mvrhel_laptop: yeah, I know. At least there's a real customer at the end of that (maybe) | 15:02.40 |
chrisl | rayjj: what's the hardware on the C board? | 15:04.00 |
rayjj | mvrhel_laptop: but we don't have actual printer code from them -- just the 'reference' examples, which should be enough for benchmarking. But the missing parts are frustrating | 15:04.03 |
| a DC1589 chip (SOC) and all kinds of peripherals -- ethernet, fax/phone, device USB(2), USB host (1), speaker/mic plugs, 16 buttons what looks like SODIMM for expansion DRAM (but no info on that) | 15:06.42 |
| chrisl: and what might be an SD card slot (also no info) | 15:07.26 |
chrisl | Presumably you don't need to worry about most of the peripheral stuff? | 15:07.48 |
rayjj | the biggest lack is support for FS (only ROM/RAM) -- I wanted to be able to use the SD card :-( | 15:08.28 |
chrisl | Ah, that is a pain :-( | 15:08.53 |
rayjj | and we don't really need their ROM/RAM FS since we have our own | 15:09.16 |
| like I said -- if it was linux, it'd be done weeks ago | 15:09.47 |
| even their so called TCP/IP stack is rudimentary | 15:10.16 |
chrisl | That's pretty unforgivable these days | 15:10.41 |
kens | rayjj the problem seems to be that we set the ProcessColorModel to DeviceGray, which causes pdfwrite to set some internal state in its device structure (color_info structure). THere's nothing apparently wrong with this. We then proceed until we get the .installpagedevice, which uses the saved dictionary. At this point, we have not apparently restored back, or otherwise altered, the ProcessColorModel to RGB. This causes the rangecheck error. | 15:30.35 |
| rayjj for what its worth, this is not a new error. 9.00 behvaes in the same way for me, though it doesn't throw a memory error. | 15:32.49 |
| Looks like this is two problems :-( | 15:33.37 |
hyper_ch | hi chrisl | 17:51.57 |
chrisl | hyper_ch: hi there | 18:00.50 |
hyper_ch | chrisl: btw, made a bug report about that :) | 18:15.13 |
| https://www.libreoffice.org/bugzilla/show_bug.cgi?id=83108 | 18:15.35 |
chrisl | Cool. but as I said, past experience says they won't be interested :-( | 18:16.29 |
hyper_ch | but the report is filed :) | 18:18.41 |
| there's actually a libreoffice conference coming up beginning of september not too far away from my place | 18:19.07 |
| maybe I should go there, abduct a LO dev and make him fix that ;) | 18:19.38 |
chrisl | It might be the only way to get it done..... | 18:20.03 |
hyper_ch | it's only about 180km away from me | 18:21.05 |
| Forward 1 day (to 2014/08/27)>>> | |