| <<<Back 1 day (to 2014/04/09) | 2014/04/10 |
juschz | Hi everyone, I have a question about makefile and mupdf. I want to desactivate CJK font (Set NOCJK to skip all CJK support (this also omits embedding the CJK CMaps // pdf/pdf-fontfile.c). I try to add -DNOCJK on the CFLAGS variable (in Makefile), but this is not working. Can somebody to help me ? | 08:32.33 |
jogux_mac | Hi juschz, I'm not sure if the people that know ghostscript are around right now - but if you can hang around for a bit I'm sure someone will appear later :) | 08:38.33 |
kens | jogux, its a MuPDF question, not Ghostscript.... | 08:38.50 |
jogux_mac | oh, crikey. uh. I need a cup of tea, clearly. | 08:39.08 |
juschz | yes, but there is no mupdf irc, right ? | 08:40.07 |
jogux_mac | juschz : yes, you're in the right place | 08:40.16 |
kens | juschz : correct, we deal with MuPDF here t the moment | 08:40.20 |
| But you're a little early for the MuPDF developers | 08:40.40 |
juschz | mupdf guys are based on US ? | 08:40.53 |
kens | No, just late risers :-D | 08:41.03 |
juschz | :) | 08:41.35 |
kens | Stick aroud for a short while and someone should be online, say an hour or less | 08:41.41 |
juschz | ok, thx | 08:41.48 |
Robin_Watts | XCFLAGS, not CFLAGS. | 09:07.51 |
juschz | hi | 09:21.41 |
ghostbot | hola | 09:21.41 |
juschz | @Robin_watts : like this : XCFLAGS := -DNOCJK | 09:23.56 |
| CFLAGS += $(XCFLAGS) -Iinclude -I$(GEN) | 09:23.56 |
| ? | 09:24.02 |
Robin_Watts | Hmm. Just looking. | 09:24.35 |
| make XCFLAGS="-DNOCJK" | 09:25.45 |
| Or if you want to hardwire it, add a CFLAGS += DNOCJK to the top of Makefile | 09:27.40 |
| Or if you want to hardwire it, add a CFLAGS += -DNOCJK to the top of Makefile | 09:27.51 |
juschz | I still have GEN generated/gen_font_cjk.h | 09:30.37 |
| GEN generated/gen_font_cjk_full.h | 09:30.38 |
Robin_Watts | yes. | 09:31.05 |
juschz | is it normal ? | 09:31.25 |
Robin_Watts | yes. | 09:31.30 |
| I'd hope that it should be dropped by the linker though. | 09:31.42 |
chrisl | kens: bug 693221 - the file is altona_technical_1v2_x3.pdf | 09:32.28 |
kens | aha thanks chrisl I'll stick tha into the thread | 09:32.40 |
chrisl | It's in the command line in comment #1 | 09:32.50 |
kens | D'oh | 09:32.57 |
chrisl | Oh, not comment #1, the "Description" entry.... | 09:33.16 |
kens | Yeah its OK I see it now | 09:33.24 |
juschz | @Robin_Watts After ndk compilation, my lib still have CJK Fonts | 09:33.42 |
| (we do not use CJK, I just try to reduce final .so size) | 09:34.20 |
kens | chrisl, when I run that file on windows, to ppmraw it crashes :-( | 09:37.53 |
chrisl | kens: Hmm, I'll try it on Linux..... | 09:38.34 |
kens | No its OK | 09:38.38 |
| type in OutptuFile | 09:38.43 |
chrisl | And that caused it to crash??? | 09:38.57 |
kens | Seems ot yes | 09:39.04 |
| I had a capital I in OutputFile | 09:39.24 |
Robin_Watts | juschz: Load pdf-fontfile.c into an editor. | 09:44.40 |
juschz | yes | 09:44.51 |
Robin_Watts | Just after the first comments you'll find #ifdef NOCJK | 09:44.57 |
| just after that (before the #endif) insert your favorite expletive and then rebuild. | 09:45.26 |
| That will tell us if the NOCJK flag is making it to the compiler or not. | 09:45.54 |
juschz | CC build/debug/pdf/pdf-fontfile.o | 09:50.00 |
| source/pdf/pdf-fontfile.c:19:1: error: expected identifier or '(' | 09:50.00 |
| "HELLOWORLD" | 09:50.01 |
| with a make XCFLAGS="-DNOCJK" | 09:50.12 |
Robin_Watts | Perfect. | 09:50.17 |
| So that NOCKJ is being set. | 09:50.30 |
| Sorry. So NOCJK is being set. | 09:50.45 |
| Which means NOCJKFONT is being set. | 09:50.53 |
| Which means neither gen_font_cjk_full.h or gen_font_cjk.h will be being included. | 09:51.13 |
juschz | ok, thx | 09:56.47 |
| Seems to be ok. | 09:57.02 |
Robin_Watts | paulgardiner, tor8: I bought a pair of the torphones from amazon. Got a second hand pair for 145. | 10:56.01 |
tor8 | Robin_Watts: what made you change your mind? | 10:56.19 |
Robin_Watts | tor8: I tried paulgardiners pair on the plane last week with a better mp3 player. | 10:56.37 |
| (or at least one that gave more volume) | 10:56.44 |
| They are so vastly much better for noise cancelling, and the sound is good enough for my cloth ears. | 10:57.17 |
paulgardiner | 145. Excellent | 10:57.19 |
tor8 | Robin_Watts: yeah. no point in good sound if all you hear is airplane noise anyway :) | 10:57.40 |
| Robin_Watts: kens: I take it you've seen the Chaos Reborn kickstarter? | 10:57.59 |
kens | No ? | 10:58.06 |
tor8 | https://www.kickstarter.com/projects/1206403106/chaos-reborn-from-the-creator-of-the-original-x-co | 10:58.22 |
Robin_Watts | tor8: The big reservation I had was the thinness of the cable, but paul points out that it's replacable, so that's fine. | 10:58.23 |
kens | Hmm, intriguing | 10:58.54 |
| I can barely remember the original | 10:59.32 |
tor8 | kens: then I won't ask if the original is worth tracking down :) | 11:05.11 |
kens | It might be, it wa a *very* long time ago and I could easily be confusing it with another game | 11:05.40 |
| Ah the prototype is online too | 11:10.32 |
Robin_Watts | https://www.kickstarter.com/projects/599698621/prototypo-streamlining-font-creation?ref=category | 11:12.47 |
chrisl | Hmm, I'm not sure I'm keen on yet another way to create (probably) crappy fonts | 11:16.36 |
tor8 | could be useful for creating substitute fonts, but that's about it I think | 11:17.10 |
| looks like a multiple-master / metafont hybrid of sorts | 11:18.16 |
| and we all know how good the computer modern set of fonts is, right? | 11:18.34 |
kens | :-) | 11:19.05 |
| Lunch time, back in a bit | 11:19.20 |
chrisl | I did notice this the other week: http://www.adobe.com/devnet/opentype/afdko.html | 11:20.26 |
tor8 | chrisl: yeah, I think I saw that used in the source sans/code pro open source font releases somewhere | 11:21.43 |
Robin_Watts | pops helen into station. I have to buy a new light fitment for downstairs too. | 11:37.49 |
ray_laptop | that's enough staring at the screen for today. Finished review of 1400 / 2692 files. Have 7 files to work on. Going to take a break from mind numbing review and look at these issues today. | 13:28.03 |
kens | Seems reasonable to me | 13:28.16 |
henrys | kens: do you know off the top of your head where/how pdf handles UCR/BG stuff ⦠looking at the pxlcolor bug. | 14:06.16 |
| ? | 14:06.17 |
| s/pdf/pdfwrite | 14:06.28 |
kens | pdfwrite doesn't 'handle' it exactly | 14:06.40 |
| It just copies it from the graphics state of the PostScript to the PDF graphics state | 14:06.58 |
| WHich isn't going to work for PXL I would imagine | 14:07.21 |
henrys | kens: nope | 14:07.38 |
ray_laptop | why not -- doesn't the pxl update the graphics state ? | 14:11.00 |
kens | ray_laptop : I assume it doesn't have a UCR/B in its graphics state, so it needs to modify the colours rather than emit a GS | 14:11.25 |
chrisl | PCL/PXL does lots of stuff without bothering to tell the graphics library :-( | 14:11.32 |
ray_laptop | oh, sorry -- the pxlcolor device | 14:11.36 |
chrisl | Oh, right.... | 14:11.49 |
kens | pdfwrite just sticks a UCR/BG in the PDF graphics state | 14:11.51 |
henrys | chrisl: not input, output | 14:11.53 |
ray_laptop | henrys: sorry, I was confused, too | 14:12.10 |
henrys | I don't think I've ever seen a customer bug against pxlcolor | 14:12.34 |
chrisl | That might mean rendering to get the "right" result..... | 14:12.50 |
kens | THis is a customer bug ? | 14:12.51 |
henrys | kens: 870 | 14:13.12 |
kens | chrisl it should be possible to do by 'applying' the UCR/BG (in Disitller terms) | 14:13.13 |
| wow, that is a surprise | 14:13.19 |
chrisl | kens: I wouldn't have thought that would work reliably | 14:13.54 |
kens | chrisl why not ? | 14:14.04 |
chrisl | Mostly due to images | 14:14.30 |
kens | well you have to convert the images too | 14:14.41 |
ray_laptop | henrys: bug tracker shows 14 pxlcolor customer bugs (5 were enhancements). Only bug 695129 is open, however | 14:15.43 |
kens | But this only matters when converting RGB to CMYK, so presumably the colorus are *already* being converted in some fashion | 14:16.05 |
| You just need to apply the UCR/BG at that time. | 14:16.28 |
chrisl | Hmm, of just output RGB in the PXL..... | 14:16.45 |
| s/of/or | 14:16.58 |
kens | well yes, that would make more sense perhaps. I don;t recall the bug exactly (even though I think I commented on it.....) | 14:17.11 |
| Goldish memory strikes again | 14:17.20 |
ray_laptop | customers that seem to report pxlcolor issues are 130, 351, 870 (and 710 who is terminated) | 14:20.25 |
kens | wonders if the UCR/BG is a red herring | 14:21.07 |
ray_laptop | there weren't really 14 pxlcolor bugs -- I just searched for pxlcolor in a comment and there were 5 false positives | 14:21.08 |
| kens: or is that a black herring ? | 14:21.28 |
kens | I guess maybe it is now :-) | 14:21.38 |
| Ah henrys I think its the transfer function, niot the UCR/BG you want to look into | 14:22.53 |
henrys | both would be a problem for pxlcolor | 14:24.19 |
kens | Well, the UCR/BG is only required when converting RGB to CMYK, does pxlcolor do CMYK output ? | 14:24.44 |
henrys | kens:oh right that is true. | 14:25.33 |
kens | H,, I'm not sure its the transfer function either, I just tried erasing it and it made no difference | 14:29.37 |
henrys | kens:and it seems to work fine in xpdf contrary to Hin-Tak's comment. | 14:32.24 |
| oh wait no it doesn't. | 14:32.41 |
kens | Could be xpdf supports transfer functions now, I'd try it in MuPDF | 14:32.48 |
henrys | it is broken in xpdf sorry | 14:33.35 |
kens | Its an unpleasantly complex PDF file :-( | 14:33.36 |
| OK finally foudn it | 14:35.54 |
henrys | kens:yes just looking at the -dPDFDEBUG messs | 14:35.59 |
kens | It is transfer functiosn, if I remove them from teh PDF file Acrobat renders the file with the black bits | 14:36.11 |
| I can give you a simpler file if you want henrys | 14:36.27 |
henrys | kens:that would be great | 14:36.38 |
kens | OK let me tidy this up a bit and I'll amil it | 14:36.50 |
henrys | kens: might as well attach it to the bug - better to have a simple file in the cluster tests. | 14:37.31 |
kens | OK NP | 14:37.41 |
henrys | wonders if we detect a change to the transfer function can we "punt" down to a low enough level so the graphics library applies it. | 14:38.55 |
| ray_laptop: the transfer function must be applied before copy_color, isn't it? | 14:39.36 |
paulgardiner | Is there a Photoshop expert among us? | 14:40.15 |
henrys | paulgardiner: I think michael the most expert but fire away. | 14:40.49 |
ray_laptop | paulgardiner: I use Photoshop, but not much of it | 14:42.17 |
| henrys: yes, the xfer function comes before copy_color (colors should already be munged) | 14:42.59 |
paulgardiner | I'm just looking at replacing "Picsel" with "Artifex" in the SOT artwork. I can probably do an acceptable job of it, but I was just thinking someone with experience might do it faster and better. | 14:42.59 |
ray_laptop | paulgardiner: I should be able to do that | 14:43.21 |
| where is the SOT artwork ? | 14:43.31 |
paulgardiner | resources/assets/smart-products | 14:44.32 |
Robin_Watts | paulgardiner: Do we have the original photoshop files? | 14:44.36 |
paulgardiner | Robin_Watts: yes we do. | 14:44.47 |
ray_laptop | paulgardiner: I have the epage git and there are only 'fonts' and 'certs' in resource | 14:45.18 |
paulgardiner | I had a go with GIMP, but I couldn't quite get an exact match for the font weight and the kerning wasn't great either | 14:45.31 |
kens | henrys 2 files on the bug for you, one is simply reduced, the other is the same file but with the transfer function removed so you can see the effect. | 14:45.42 |
ray_laptop | paulgardiner: so, where is 'assets' | 14:46.05 |
henrys | kens: thanks I'll look | 14:46.16 |
paulgardiner | ray_laptop: is that the old epage-only repo. We now have resources and epage combined. | 14:46.16 |
ray_laptop | OK, I'll trash that git. Where should I git it from ? | 14:46.48 |
Robin_Watts | ghostscript.com:/home/git-private/sot.git | 14:47.07 |
ray_laptop | what's the fastest way to get it locally ? | 14:48.34 |
Robin_Watts | ray_laptop: There is probably a copy on peeved. | 14:49.12 |
ray_laptop | Robin_Watts: and where would that be ? | 14:50.15 |
| (on peeved) | 14:50.20 |
Robin_Watts | ray_laptop: just checking :) | 14:50.22 |
henrys | kens:well the simplified file is correct using pxlcolor, so something is awry | 14:50.30 |
ray_laptop | Robin_Watts: OK. thanks. | 14:50.32 |
Robin_Watts | ray_laptop: /home/robin/PicselCVS/sot.git | 14:50.42 |
ray_laptop | In the meantime, I'm cloning the sot.git | 14:50.56 |
henrys | kens:oh hold on, that may be wrong | 14:51.19 |
ray_laptop | but that'll probably take a while | 14:51.20 |
| 1.4 million objects | 14:51.49 |
| uh, paulgardiner: there are 26 .psd files there | 14:54.58 |
henrys | kens:okay reduced.pdf does reproduce the same issue, thanks. | 14:55.22 |
kens | :-) glad its wrong I guess .... | 14:55.34 |
paulgardiner | ray_laptop: I think I can cut that down to just 6 or 7. | 15:01.49 |
Robin_Watts | ray_laptop: Do you have CS3 or later ? | 15:02.51 |
ray_laptop | paulgardiner: I don't have the font required "DIN Regular" | 15:05.05 |
| Robin_Watts: yes | 15:05.12 |
| I have CS3 | 15:05.16 |
paulgardiner | ray_laptop: if I can just get hold of a copy of CS3 I may not need to waste anyone else's time with this. I imagine the alterations, although possibly laborious, will be easy enough. | 15:05.58 |
ray_laptop | I can see the text layers for PICSEL, SMART VIEWER and SMART OFFICE | 15:06.16 |
paulgardiner | ray_laptop: I could see those in GIMP, but only as image data. I couldn't see info on the font | 15:07.14 |
ray_laptop | I can see it as text, but I don't have the DIN Regular font, so if I change it, it will be to a different font | 15:08.04 |
Robin_Watts | ray_laptop: We may have that. | 15:09.02 |
| On points, in /mnt/picsel/dataserver/central/transfer/alexms/fonts/Fonts | 15:10.01 |
| there are many fonts, including DIN ones. | 15:10.08 |
| dataserver/central/transfer/IainFrame/Desktop files/fonts/Adobe Font Folio 11/Western Fonts has even more | 15:11.08 |
ray_laptop | hmm... I accept the "If you change this font substitution will occur", but then it looks the same. | 15:12.55 |
| I'll go ahead and get the DIN Regular and re-try opening it | 15:13.21 |
jogux_mac | ray_laptop : I suspect some of the 'text' is paths :-( | 15:15.01 |
ray_laptop | jogux: not for those layers. | 15:15.46 |
| in the Icons, yes | 15:15.56 |
henrys | kens:yeah simply punting on the high level image fixes it. Now to figure out how to detect it. | 15:16.00 |
jogux_mac | that's good at least. | 15:16.00 |
ray_laptop | it looks like the product names and company name are made to be altered | 15:16.35 |
| oh, "only" 596 fonts there. | 15:17.40 |
kens | henrys I might be able to find out where pdfwrite detects a change in trasnfer function, but I suspect you simply want to find 'not the identity' | 15:19.18 |
henrys | kens:I asked hin-tak if he want s to follow up, he's done a lot of work in the device and may want to fool with it. | 15:20.38 |
kens | Hmm pdfwrite maintains a stack of graphics states and looks to see if the current state is the same as hte most recently stored state, if its not then it emits a change and stores the current state | 15:20.44 |
ray_laptop | I guess it won't hurt to add those to my machine (I already have 649 otf and ttf loaded on my laptop) | 15:23.00 |
Robin_Watts | ray_laptop: You can probably get away with just the DIN ones? | 15:24.08 |
ray_laptop | only 385 ttf and otf on the picsel server | 15:24.10 |
henrys | kens:I think the imager state has the info. pis->effective_transfer | 15:24.17 |
ray_laptop | Robin_Watts: I know, but you can never have too many fonts :-) | 15:24.28 |
kens | henrys I think checking for pis->set_transfer.red green or blue not being a NULL pointer may be enough | 15:24.52 |
| At least that's what I see here | 15:25.13 |
henrys | kens:makes sense we definitely want to avoid false positives though the high level image is important performance wise | 15:25.46 |
kens | Well, you could apply the transfer and see if its the identity, but that would be slow. I think just checking to see if we have a transfer function set is probably enough | 15:26.39 |
henrys | kens:well turned out to be easy, I thought that would be pain. I'll wait to hear back from hintak and return to guillaume (I know that's a pain) | 15:27.47 |
ray_laptop | henrys: paulgardiner: what do we want the splash screen to say instead of "PICSEL" -- "ARTIFEX" or "ARTIFEX SOFTWARE" or what ? | 15:34.41 |
| Robin_Watts: thanks. I have the fonts installed now and don't get the gripe from CS3 anymore | 15:35.22 |
henrys | ray_laptop: I would just replace picsel with artifex leaving whatever letter case exists intact. | 15:36.05 |
ray_laptop | paulgardiner: so I am good to go on changing the files. (hopefully less that 26 of them) | 15:36.06 |
paulgardiner | ray_laptop: I assumed just ARTIFEX | 15:36.07 |
ray_laptop | I think that font is *only* uppercase | 15:36.29 |
| but I could change to a different font (such as the one we use on the website and literature) | 15:37.20 |
| I can even add in the Artifex "A" logo -- I have it in vector form | 15:43.34 |
paulgardiner | ray_laptop: probably is worth changing all 26. Is that a lot of hassle? If so, it's possibly worth holding off because we've found a copy of CS3 that I might be able to use amongst the Picsel spoils. | 15:47.24 |
ray_laptop | paulgardiner: henrys: here's one for review (PNG) casper.ghostscript.com:~ray/landscape-splash-artifex.png | 15:48.58 |
| paulgardiner: it's actually reasonably quick. | 15:49.31 |
paulgardiner | oh okay great | 15:49.43 |
ray_laptop | I've just been interleaving other stuff. | 15:49.47 |
| henrys: paulgardiner: Robin_Watts: (anyone) let me know if that looks OK | 15:50.09 |
paulgardiner | ray_laptop: just grabbing it | 15:50.24 |
Robin_Watts | ray_laptop: Nice. | 15:51.20 |
ray_laptop | I put it at http://casper.ghostscript.com/~ray/landscape-splash-artifex.png as well | 15:51.24 |
Robin_Watts | We probably need the psd and png files back for each one, I would have thought. | 15:51.40 |
ray_laptop | Robin_Watts: I just pushed the png for review. I know we will want the PSD's as well | 15:52.12 |
henrys | ray_laptop: looks nice | 15:52.17 |
ray_laptop | paulgardiner: OK. give me a few minutes and I'll do the rest (that was the xga one) | 15:53.10 |
paulgardiner | ray_laptop: yeah looks great. GIMP's kerning really wasn't good | 15:53.11 |
ray_laptop | paulgardiner: you probably didin't have the right font, either | 15:53.38 |
paulgardiner | ray_laptop: no I didn't. I guess I could have looked at the xml in the psd file but didn't think of that until now | 15:54.23 |
ray_laptop | now there are layers in the PSD that are switchable for the Lite vs regular icon and the SMART VIEWER vs SMART OFFICE text | 15:54.43 |
paulgardiner | ray_laptop: we may need pngs for both, but I can definitely do that bit here once the psd files are updated | 15:55.43 |
ray_laptop | so I don't know how the splash for a particular product gets done from the PSD | 15:55.48 |
| paulgardiner: just as long as whatever you use doesn't re-do the font rendering | 15:56.16 |
| paulgardiner: do we need the 1.x and 2.x products (or is 1.x obsolete) ? | 15:57.01 |
paulgardiner | ray_laptop: yeah. pretty sure the psd holds an image in addition to the font info and GIMP just picks out the image | 15:57.08 |
Robin_Watts | paulgardiner: not sure how that works with layers. | 15:57.23 |
paulgardiner | ray_laptop: I think we just need 2.x | 15:57.25 |
Robin_Watts | presumably only the currently selected layers are rendered. | 15:57.33 |
ray_laptop | paulgardiner: well that cuts down the number of files | 15:57.45 |
Robin_Watts | certainly when I've written psd reading code before each bitmap layer is just a component. | 15:58.13 |
ray_laptop | Robin_Watts: right, so when you change the selected layers with GIMP it will (probably) have to re-render | 15:58.14 |
Robin_Watts | ray_laptop: Right. So paulgardiner will need to have CS3 working to get the pngs out, I reckon. | 15:58.35 |
paulgardiner | I think the psd files have images for each layer | 15:58.44 |
ray_laptop | unless it has each layer rendered in the PSD that GIMP can use | 15:58.53 |
paulgardiner | I was able to perform the swapping that you refer to under GIMP | 15:59.02 |
ray_laptop | doesn't know (I don't use gimp) | 15:59.08 |
| paulgardiner: OK. Well, I'll change the 2.x psd's and package them up for you | 15:59.38 |
paulgardiner | ray_laptop: excellent | 15:59.49 |
ray_laptop | paulgardiner: oops. It was taking forever to download them -- I forgot -C | 16:05.41 |
| that about triples the speed (psd's aren't compressed) | 16:06.32 |
| of course, it doesn't help that I'm cloning the sot git at the same time :-/ | 16:08.20 |
| and streaming a movie ;-) | 16:08.44 |
paulgardiner | No that is probably pushing it a bit | 16:09.51 |
jogux_mac | paulgardiner : two trivial reviews for you in my tree when you have a minute (not urgent) | 16:49.51 |
paulgardiner | jogux_mac: looks good to me | 16:53.05 |
jogux_mac | you said that about the one that broke it too ;-) | 16:53.16 |
| ta, pushed :) | 16:53.54 |
Robin_Watts | jogux_mac: So... tgv-sample-code.pl does some really funky stuff with attempting to strip down the code that it ships. | 16:55.53 |
| For each C file it runs through and removes as much conditionalised code as it can. | 16:56.08 |
| In order to do that it needs to figure out what macros are defined in system headers, and what in the source itself. | 16:56.39 |
jogux_mac | well I figured the 3,800 lines of perl had to be doing something :-S | 16:57.00 |
Robin_Watts | And in order to do *that* it calls cpp with -H -dD | 16:57.05 |
jogux_mac | ahha. | 16:57.16 |
Robin_Watts | BUT, the format of the output given by 4.7 and 4.3 differ, I think. | 16:57.24 |
jogux_mac | yay | 16:57.29 |
Robin_Watts | and that's upsetting the perl and it's thinking that all files are empty and hence ignoring them. | 16:57.49 |
jogux_mac | how do we know it was 4.3? | 16:58.03 |
Robin_Watts | Well, I just tried calling the mips cpp rather than the current wheezy one :) | 16:58.46 |
kens | Goodnight all | 16:58.49 |
Robin_Watts | and it got further. | 16:58.55 |
| kens: night. | 16:58.57 |
jogux_mac | ah :-) | 16:58.58 |
| robin_watts: I guess one step would likely be to make it fail out if it finds no matching lines. | 17:01.18 |
ray_laptop | paulgardiner: sorry -- had a phone call interruption. I'm uploading the files to casper:~ray/artifex-sot-splash.zip | 17:02.34 |
Robin_Watts | Hmm. If I leave it calling the mips one, it falls over missing picsel-config-defines.h | 17:02.39 |
| That seems fairly crucial. | 17:02.45 |
| (it gets past the cpp stage, but then dies in the build. Not sure how that works) | 17:03.09 |
ray_laptop | paulgardiner: that's the PSD's I'll leave the rest for you since you know what format you need them in, and which products are needed (selecting appropriate layers) | 17:04.00 |
| paulgardiner: I'll let you know when it's finished uploading (about 36Mb) | 17:04.38 |
Robin_Watts | jogux_mac: The code seems ready to expect certain files to be completely empty after stripping. | 17:09.17 |
| There is code to specifically cope with that case. | 17:09.28 |
jogux_mac | nods | 17:09.42 |
ray_laptop | paulgardiner: OK. uploaded. These are the 2.x files, and it turns out there are 16 files. Also I didn't change it, but the nokia-symbian landscape is named funny: portrait-landscape.psd instead of landscape-splash.psd | 17:11.19 |
paulgardiner | Great. Thanks ray_laptop. Much appreciated | 17:22.59 |
ray_laptop | paulgardiner: np. Better than staring at image comparisons, or even trying to figure out the fast thresholding vs. "normal" image diffs | 17:25.03 |
Robin_Watts | jogux_mac: Has fermis died for you? | 17:25.27 |
| seems to have died on the cluster too. | 17:25.49 |
jogux_mac | it was very slow but I got in | 17:26.12 |
henrys | tor8: just to let you know the fonts are delayed miles is hoping to get financing from a customer to help out. | 17:26.40 |
Robin_Watts | jogux_mac: Yeah, back now. | 17:27.00 |
marcosw | jogux_mac: the testdocs are on fermis now, in /mnt/ats/testdocs. I'm in the process of rsyncing them to plancks, which is why ssh access to fermis might be a bit slow. | 17:47.30 |
jogux | marcosw : excellent, thanks | 18:33.01 |
Robin_Watts | This is just so painful. I remember now why I hated doing anything with the build scripts at all. | 18:34.32 |
| So it turns out that the way the cpp is being called has killed all the standard include paths. | 18:35.11 |
| This means that the moment it hits #include <assert.h> the rest of the file is ignored. | 18:35.26 |
| The old cpp doesn't die with an error if it can't find an include file. The new one does. | 18:43.22 |
ray_laptop | my clone of the sot git finally finished. I gor 1881 errors about tags not pointing to valid objects, but git status is OK | 19:29.34 |
| s/gor/got/ | 19:30.01 |
| ugh. having to track down image position differences between fast thresholding and normal processing :-( | 19:36.53 |
jogux | ray_laptop : I think we decided the fix for those errors was to upgrade git (1.7 not good, 1.9 good, iirc) | 20:30.37 |
| robin_watts : urgh :( | 20:30.48 |
Robin_Watts | jogux: So, I think tomorrow I'm going to try to set up the build on the etch image rather than the wheezy one. | 21:06.37 |
SidneyAKS | Hi, anyone available for a little trouble shooting? I may have found a bug, or... Something. | 23:28.21 |
| Forward 1 day (to 2014/04/11)>>> | |