| | <<<Back 1 day (to 2015/07/07) | 20150708 |
| rayjj | very interesting (and somewhat concerning). The image_render_color code doesn't use a (fixed_half - fixed_epsilon) adjustment, but the image_render_mono *ONLY* does when it is the "Fast case: no skew, and not imagemask with a halftone" | 02:27.57 |
| | since it is only done in this one case, I am thinking it is wrong. | 02:28.31 |
| ao2 | hi, everyone. Does anyone know why the muPDF android app requires full network access? | 07:13.42 |
| kens | To read PDF files from URLs probably | 07:14.08 |
| ao2 | kens, I see, thanks. Maybe I'll try rebuilding the app with this permission disabled and see what happens. | 07:16.52 |
| kens | Ha, my comment collided with Chris's :-) | 07:23.30 |
| chrisl | Saying the same thing? | 07:24.01 |
| kens | Pretty much yeah | 07:24.07 |
| | Looks identical to me | 07:24.14 |
| chrisl | I await the (likely bitchy) reply...... | 07:24.44 |
| kens | Oh good grief, Aaron is setting -r72 to pdfwrite..... | 07:24.57 |
| chrisl | I noticed that - goodness knows why | 07:25.53 |
| kens | Idiocy I suspect | 07:26.01 |
| | I'm not even sure which part of thr (large and complex) page he' complaining about | 07:27.06 |
| chrisl | I'd ignore it for now - Marcos seems to be dealing with it well enough | 07:28.46 |
| kens | Oh it looks like he sees stepping in the shading. Probably because its not a shading but is calculated in PostScript at a guess. | 07:28.47 |
| | Yeah I was just curious | 07:29.01 |
| | Oddly it looks like its a JPEG, it has artefacts in it, so it may well be just that. Not surprising since they are using PDFSETTINGS and not overriding the colorimage filter | 07:30.15 |
| | Looks like the real problem is that he's downsampling images above a certain threshold (that's in the /printer settings). If he runs at high resolution, then the threshold isn't reached and no downsampling takes place. If he runs at low resolution it is, and so it downsamples. Surprisingly downsampled images are not the same as the originals...... | 07:34.57 |
| wilr | chrisl: may I ask permission to have access to the details of bug #696070 which was just marked as duplicate of 696041 (the one I reported) ? | 08:24.48 |
| kens | I'm not sure that's possible | 08:25.07 |
| | You would have to be the bug reporter, or a member of the Artifex staff | 08:25.24 |
| | Possibly we could add you to the CC list if you supply a mail address | 08:25.39 |
| kens | goes to look at other bug | 08:26.01 |
| | Try now | 08:26.53 |
| wilr | kens: it works, thanks ! | 08:27.41 |
| kens | Ah, never done that before, nce ot know it works | 08:27.50 |
| chrisl | wilr: Ah, I was about to e-mail you the details.... never mind, then! | 08:34.11 |
| wilr | chrisl: no problem, now I think I have enough information to write the advisory and publish it | 08:35.36 |
| sebras | does anyone know if there is a setting in pdf that enables write permissions in a form? what should I be looking for? | 08:58.07 |
| kens | What kind of form ? Form XObject ? XFA form ? PDF form ? SOmething else ? | 08:58.36 |
| | Should have said AcroForms for that last one | 09:01.03 |
| | AcroForm fields can have a flag which prevents users changing the value | 09:01.58 |
| sebras | kens: I'm not sure. https://www.forsakringskassan.se/wps/wcm/connect/70de9124-b003-4c0d-addd-3af1bf991aa1/fk7461_014_f_001.pdf?MOD=AJPERES | 09:06.17 |
| | kens: some fields, e.g. the one labeled "Du som ansöker" in the upper left on page 1 I can enter text into. | 09:06.59 |
| | kens: others, e.g. the 2nd line of "Uppgifter om arbete" in the upper left on page 2 I can not enter text into. | 09:07.34 |
| kens | Its an AcroForm with XFA content, so a hybrid | 09:09.07 |
| sebras | kens: ok, teach me. how do I see this? | 09:10.22 |
| kens | I decompressed it with MuPDF then searched for AcroForm | 09:10.36 |
| sebras | kens: ah! /AcroForm in the root. | 09:10.53 |
| kens | And teh AxcroForm object has a /XFA key | 09:11.10 |
| | Most of the fields look like they are AcroFOrm fields | 09:11.59 |
| sebras | kens: does XFA describe whether individual fields are enabled/disabled? | 09:12.04 |
| kens | I don't understand XFA | 09:12.13 |
| | Because its XML and incomprehensible | 09:12.22 |
| sebras | :) | 09:12.31 |
| sebras | unfortunately learned xslt recently | 09:12.42 |
| | kens: so basically it is object 102 and its descendants that describe the fields on all pages in the document? | 09:13.07 |
| kens | It look s to me like the fields are regular AcroForm fields | 09:13.21 |
| | Object 108 I thnk | 09:13.36 |
| | Is there a particular field you are having a problem with ? | 09:14.18 |
| sebras | kens: ah, yes. I did pdfclean, let me go back to the original pdf. | 09:14.33 |
| | kens: yes on the second page in the lower left corner there is a field with the caption "Datum" which is write-protected for whatever reason. | 09:15.12 |
| kens | Let me look | 09:15.23 |
| sebras | there are others too, but I think they are all the same. | 09:15.26 |
| kens | Its a date field in Acrobat | 09:15.40 |
| | There's a button and if I press it then I get a calendar | 09:16.02 |
| | Its also running Javascript, and I broke it :-) | 09:16.28 |
| sebras | oh! | 09:16.42 |
| | I tried evince too and I can't enter any text there but I see the field. | 09:16.57 |
| kens | Type a non-date intot he field, and it pops up a JavaScript dialog telling me I messed up. Click OK and it pops straight back up agaiun..... | 09:17.15 |
| sebras | the same goes for the field in the lower right labeled "Telefon kvällstid, även riktnummer" | 09:17.26 |
| kens | I thnktht field is an XFA field and it runs JavaSAcript, so very possibly you can't edit it without Acrobat otr a very full XFA compliant editor | 09:18.36 |
| | I can edit the 'Tefon' field | 09:19.49 |
| | orry Telefon | 09:19.54 |
| sebras | kens: oh.. that's surprising. | 09:20.20 |
| kens | Well I'm using Acrobat Pro | 09:20.30 |
| sebras | kens: hm. maybe I need to fire up acroread. :( | 09:20.31 |
| kens | Acrorfead won't do | 09:20.37 |
| sebras | kens: (btw, Telefon means Phone) | 09:20.41 |
| kens | At least, I think it won't | 09:20.43 |
| | sebras, yes I know, same in German | 09:20.51 |
| rayjj | I am _so_ done with fast thresholding. The image code paths in gs are hideous. | 14:32.29 |
| mvrhel_laptop | rayj: I thought you were done | 14:49.20 |
| | rayjj^^ | 14:49.24 |
| sebras | kens: I had to resort to using acroread for windows in the end. the acroread that adobe supplied for linux is 32-bit only (and I didn't want to install a 32-bit environment only for this reason). | 15:03.56 |
| kens | has no clue, I'm somewhat surprised that Reader allows it. | 15:04.58 |
| sebras | kens: there were a few datefields where a calendar popped up, and I can understand that neither mupdf nor evince manages to handle those (yet). | 15:08.06 |
| | kens: but there were several other fields that I _could_ edit in acroread but that I was unable to edit in mupdf/evince. | 15:08.31 |
| | I need to dig into this further I guess. :-/ | 15:08.42 |
| kens | I'm not sure if those are JavaScript or XFA, either way MuPDF might eventually get the ability to handle it | 15:08.47 |
| | I'm not sure evince handles XFA | 15:09.07 |
| sebras | kens: yes, I think mupdf should be able to do so. | 15:09.12 |
| | kens: aha, maybe that was the cause in evince then. | 15:09.35 |
| Robin_Watts | rayjj: I think you've misunderstood Zoltans request. | 15:16.33 |
| | I think he wants to be able to take a PDF stream (e.g. 100 0 0 100 0 0 cm 100 100 m 200 200 l ...) etc and then add debugging to it (100 0 0 100 0 0 cm (foo) == 100 100 m (bar) == 200 200 l (baz) == ...) | 15:18.03 |
| rayjj | Robin_Watts: yeah, I figured that was what he wanted, but I wasn't willing to do that | 15:19.35 |
| Robin_Watts | rayjj: "I understand what you want, and we can't help you." | 15:21.00 |
| rayjj | Robin_Watts: I just figured I would give him some hints on debugging from PDF's -- it'll probably help | 15:27.15 |
| | mvrhel_laptop: I got the transfer function sorted out, but I was still getting lots of diffs. I tracked those to dda differences resulting in obo pixel differences LOTS of places. That's when I found that the image code uses different methods for scaling (one place with a fixed_half-fixed_epsilon fudge, others without) | 15:29.29 |
| | have to run an errand. bbiab | 15:29.53 |
| mvrhel_laptop | :( | 15:29.53 |
| rayjj | mvrhel_laptop: and 'fixing' it caused a segfault to crop up. AND I have tests/Ghent_V3.0/052_Font_report.pdf that *looks* like it might be a transfer function issue still (maybe) | 15:31.42 |
| mvrhel_laptop | that does sound like a nightmare | 15:32.08 |
| rayjj | mvrhel_laptop: I often wake up thinking about it -- I guess that qualifies as a nightmare :-) | 15:32.43 |
| | hmm... I don't see the difference on tests/Ghent_V3.0/052_Font_report.pdf locally. :-( | 16:13.12 |
| henrys | errand be back in a bit | 16:35.41 |
| Robin_Watts | I think that's gs_error_and_be_back_in_a_bit these days. | 16:36.58 |
| jogux | hmm. we should possibly get Miles to install an ad blocker sometime :) | 16:59.28 |
| Robin_Watts | ? | 17:05.23 |
| jogux | Robin_Watts: I sent him a link to a new story and got back a comment "Did you notice the plastic surgery disaster picture? Ouch...." | 17:23.51 |
| | s/new/news/ | 17:23.56 |
| Robin_Watts | ah :) | 17:33.34 |
| | I survive without an ad blocker. | 17:33.43 |
| henrys | jogux: when I first read that I thought spam is getting more sophisticated they've managed to inject stuff in email, but I guess Miles said it. | 18:18.12 |
| jogux | :) | 18:27.30 |
| henrys | jogux: message on skype | 18:35.52 |
| jogux | henrys: ah, yep. Will proceed then. | 18:40.57 |
| rayjj | found the cause of the segfault with the fast ht code changes. It also explains a "I don't understand why" from RJW in gxht_thresh.c | 19:51.55 |
| | yet another cluster run ... | 19:52.35 |
| | NB, I still haven't explained the difference on tests/Ghent_V3.0/052_Font_report.pdf that looks like a shading difference that I can't reproduce locally | 19:54.00 |
| | Forward 1 day (to 2015/07/09)>>> | |
IRC Logs