| | <<<Back 1 day (to 2013/09/11) | 2013/09/12 |
| kens | thinks chrisl is feeling grumpy this morning, based on gs-devel replies | 07:27.07 |
| chrisl | Well, not too happy about: "I'm too lazy/stoopid to use google, do my work for me....." kind of questions. | 07:33.06 |
| kens | chrisl or read teh documentation.... | 07:49.42 |
| chrisl | kens: yeh, I very nearly did just put RTFM | 07:51.42 |
| kens | Its always tempting :-) | 07:51.51 |
| | THough it seems he thinks I was somewhat abrupt too, from his tone | 07:52.19 |
| chrisl | Well, "I don't know what Ghostscript is", FFS! | 07:52.35 |
| kens | Yep, its not exactly hard to find out.... | 07:52.52 |
| | I was more amused by the fact that he knew about AutoRotatePages, but didn't think to look for the same thing in our documentation | 07:53.25 |
| | And their 'tool tech guy' says they can't change anything. Really ? Have you conidered looking for a new one ? This one is clearly not fit for purpose | 07:54.10 |
| chrisl | He clearly hadn't thought to look at our documentation at all.... | 07:55.04 |
| Tomek_ | Hello. I would like to ask about MuPdf compilation. Do think that it is possible to statically compile MuPdf with X11, so that it forms a self-contained package? Thanks in advance for the information | 08:00.05 |
| kens | Tomek_ : you are a bit early for the MuPDF developers, if you can hang around for an hour or so someone may be able to answer your question. The short answer I suspect is that it can be done, but why would you want to ? | 08:01.04 |
| Tomek_ | kens: thanks for letting me know; I am working with a research tool, that interprets LLVM bitcode; in order to run applications with this tool I need to first compile them, ideally statically to have everything inside the bitcode file | 08:03.04 |
| kens | hmm, well I'm not a MuPDF developer, I can't see why it wouldn't be possible, but you want Robin_Watts or Tor. Neither are online yet (robin's computer is logged in, but I don't think he's there) | 08:04.06 |
| Tomek_ | ok, thank you very much for the information - I will try again later :) | 08:05.30 |
| | have a nice day | 08:05.37 |
| kens | You too. By the way, the channel is logged so if you can't hang around, you can check the logs for a delayed reply | 08:06.00 |
| | http://ghostscript.com/irclogs/current.html | 08:06.13 |
| Tomek_ | that's great - thanks! | 08:06.22 |
| | I thought I will add some more details to my question in this case: I compiled libX11 and libXcb, but it seems that libX11 should be statically compiled with libXcb, as it has some symbols missing from libxcb. I was wondering whether that chain of libraries dependencies might be longer and maybe because of that what I try to do is impossible | 08:09.14 |
| chrisl | Tomek_: you will need all the dependencies which, for X, is going to be quite large. It's also quite hard to get glibc to link statically. | 08:11.17 |
| Tomek_ | libc should not be a problem - I use a prepared version of uclibc | 08:11.55 |
| kens | I wonder if it might be easier to have MuPDF not compile using X (if a disply device is not required) | 08:12.01 |
| Tomek_ | is it possible to load PDF file without displaying it ? | 08:12.23 |
| kens | Tomek_ : certainly, but how useful that is, depends on what you want to do with it :-) | 08:12.45 |
| Tomek_ | essentially I am working on document recovery application; what I would like to do is to take a document that I know is causing crash / overflow in the application (some older release) and recovery it; for that I only need to know whether the document loaded properly | 08:13.28 |
| kens | Wow, a low resolution cluster run completes in 7 minutes now | 08:13.41 |
| chrisl | Tomek_: then mudraw would probably do everything you need - converting the PDF to an image file. | 08:14.24 |
| kens | Tomek_ : it should be entirely possible to do that without displaying anything. You would want to render all the pages (in case a page is broken) but you could do that to an itnernal bitmap and then throw it away | 08:14.24 |
| Tomek_ | and I could do that with mudraw ? | 08:14.51 |
| chrisl | Tomek_: I still don't see why you want to statically like with X, though...... | 08:15.29 |
| kens | Tomek_ : You need to bear in mind that MuPDF is a library, the app that everyone thinks of as 'MuPDF@ is just a demo | 08:15.33 |
| chrisl | s/like/link | 08:15.36 |
| Tomek_ | chrisl: it is a requirement originating from the architecture of the testing / execution tool that I use | 08:16.27 |
| | if I can load a file without displaying a window, then probably I will not need to link with X | 08:17.01 |
| chrisl | mudraw will do just that - like I said, converting the PDF to an image file (PPM, PNG etc) | 08:17.38 |
| Tomek_ | from what you say I understand that the document loading logic is the save when I open a file with mudraw and with mupdf | 08:18.08 |
| chrisl | No need for the pain of building X11 and it's dependencies, which I think would be horrid | 08:18.10 |
| Tomek_ | that's great - and I agree, it seems that building X will be "non trivial" ;) | 08:18.37 |
| kens | Tomek_ : yes, the loading logic is the same, however I *would* strongly suggest that you render wach page, as that will tell you if any page(s) have problems | 08:18.46 |
| | s/wach/each/ | 08:18.59 |
| Tomek_ | kens, chrisl: Thanks a lot - I will check that | 08:19.18 |
| chrisl | Tomek_: np | 08:19.25 |
| Tomek_ | that's really helpful for me | 08:19.29 |
| kens | chrisl did someone change the makefiles ? I just did a rebase and cluster push and I'm getting *many* 'extra dependency' warnings... | 09:06.57 |
| chrisl | kens: I'm not aware of anything | 09:07.28 |
| kens | :-( | 09:07.36 |
| chrisl | Oh, marcos changed lib.mak..... | 09:08.03 |
| kens | Umm, I wonder if that could be it | 09:08.22 |
| chrisl | Seems you left a space out of your commit - can't see that would make much difference, though | 09:08.46 |
| kens | Well no that seems reasonable | 09:08.55 |
| | I just got: | 09:09.22 |
| | sa85d.$(OBJ): extra dependency on $(MAKEDIRS) | 09:09.22 |
| | saes.$(OBJ): extra dependency on $(MAKEDIRS) | 09:09.23 |
| | sarc4.$(OBJ): extra dependency on $(MAKEDIRS) | 09:09.23 |
| | sbcp.$(OBJ): extra dependency on $(MAKEDIRS) | 09:09.23 |
| | scantab.$(OBJ): extra dependency on $(MAKEDIRS) | 09:09.23 |
| | scfd.$(OBJ): extra dependency on $(MAKEDIRS) | 09:09.23 |
| | scfdtab.$(OBJ): extra dependency on $(MAKEDIRS) | 09:09.24 |
| | scfe.$(OBJ): extra dependency on $(MAKEDIRS) | 09:09.24 |
| | scfetab.$(OBJ): extra dependency on $(MAKEDIRS) | 09:09.25 |
| | scfparam.$(OBJ): extra dependency on $(MAKEDIRS) | 09:09.25 |
| | sdcparam.$(OBJ): extra dependency on $(MAKEDIRS) | 09:09.26 |
| | sdctc.$(OBJ): extra dependency on $(MAKEDIRS) | 09:09.26 |
| | sdctd_0.$(OBJ): extra dependency on $(gdebug_h) | 09:09.27 |
| chrisl | I can only guess that marcos's change tripped up the code that filters the existing, known warnings | 09:10.08 |
| kens | ah, I guess that could do it, that hadn't occured to me | 09:10.26 |
| chrisl | It happens every now and then | 09:10.58 |
| kens | I'll just ignore it then :-) | 09:11.12 |
| chrisl | Good plan.... | 09:11.32 |
| | kens: ping | 10:15.07 |
| kens | pong | 10:15.11 |
| chrisl | Did you settle on the Delta flights? | 10:15.23 |
| kens | Not settled yet | 10:15.29 |
| | But probably | 10:15.33 |
| chrisl | I couldn't find the same return flights as you had | 10:16.02 |
| kens | I'm trying to find out if I can upgrade the Virgin legs to PE. I don't believe the Maui legs have it, so that will be extra legroom if I can get it | 10:16.07 |
| | Hmm, I can't think why you wouldn't find the flights | 10:16.49 |
| | let me check again | 10:16.54 |
| chrisl | Same from Maui to JFK, but different JFK to LHR | 10:16.56 |
| kens | boggles | 10:17.06 |
| | chrisl you are correct, that no longer shows up for me either | 10:19.29 |
| | Obviously I should have booked it yesterday :( | 10:19.39 |
| chrisl | But there is one *almost* the same | 10:19.46 |
| kens | I don't see any with a Virgin codeshare | 10:19.58 |
| chrisl | No, possibly not :-( | 10:20.16 |
| kens | THere is a 21.5 hour one | 10:20.16 |
| | I wanted Virgin so I could upgrade, but I guess if its only from JFK I can live with it | 10:20.37 |
| chrisl | Yeh, last leg is DL 401 | 10:20.42 |
| kens | THat's what I see yes | 10:20.54 |
| chrisl | I take it the 1.5 hour connection isn't a problem? | 10:21.43 |
| kens | Well I should hope not, no immigration an I won't be carrying checked luggage | 10:22.04 |
| | tor7 you there ? | 10:22.58 |
| chrisl | Hrm, and now I don't see the 21.5 hour one :-( | 10:25.54 |
| kens | chrisl either they keep changin the flights or their web site is desperately flaky | 10:28.59 |
| | the 21.5 hour one came up for me | 10:31.19 |
| chrisl | Yep, it's back now for me, too. I think I hate this....... | 10:31.37 |
| kens | I reached that point on Sunday | 10:31.52 |
| | 10 minutes on hold to Virgin and now the phone just rings...... | 10:32.57 |
| | and when they finally answer 'you'll have to call Delta' :-( | 10:35.39 |
| | If tor is OK wiht it I'm going to book these flights while I still can, and hope to upgrade after the fact. | 10:36.54 |
| chrisl | So you're looking at return flights: DL 2116, DL 16 and DL 401? | 10:40.11 |
| kens | Probably, I closed the browser tab. It was OGG->LAX->JFK->LHR arriving at something like 5:35 am | 10:40.53 |
| | and DL 401 was the last hop | 10:41.00 |
| chrisl | Okay, so I'm looking at flying out earlier than you guys, but flying back with you | 10:41.48 |
| kens | OK no problem there. | 10:41.57 |
| | But it would be good if I could contact Tor :-( | 10:42.10 |
| chrisl | And given I manage 5 days in Miami with just carry-on luggage, I can probably manage that for Maui, too..... | 10:42.43 |
| kens | I would think so, yes | 10:42.52 |
| chrisl | I'm happy with being scruffy | 10:43.09 |
| kens | plenty shorts and T-shirts I guess | 10:43.10 |
| | OK I think I've removed all the large array and string stuff. Just need to check it all still works..... | 10:43.49 |
| chrisl | Yep. I wonder if Delta allows carry-on + laptop like Virgin does..... | 10:44.08 |
| kens | I htink most airlines do, and US carriers tend to be laxer | 10:44.28 |
| chrisl | Drat, beaten to the cluster again :-( | 10:56.09 |
| kens | :-) | 10:56.14 |
| | I've just removed all the 'larray, lstring' stuff and I want to make sure it all still works | 10:56.32 |
| | Before I commit it. | 10:56.38 |
| | It'll be 5 minutes or so, its a lowres run | 10:57.10 |
| tor7 | kens: I'm here now. | 11:15.20 |
| kens | Aha, welcome back tor7 | 11:15.31 |
| | I've not found anything better than the Delta flights, and as chrisl point sout the JFK return one with Virgin has now disappeared :-( | 11:15.56 |
| tor7 | rats :( | 11:16.06 |
| | still, PE on the long leg out is the only one that *really* matters | 11:16.19 |
| kens | So I'd like to go ahead and book those flights now, before it gets any worse. Virgin say I have to ask Delta about upgrading to PE | 11:16.25 |
| | But I can't believe that'll be a problem. | 11:16.38 |
| | SO, if you are OK with it I'd like to book the flights asap | 11:17.01 |
| tor7 | kens: I sure hope it won't be a problem! | 11:17.11 |
| kens | has fingers crossed.... | 11:17.21 |
| | But worst case I'msure an upgrade could be done at the airport | 11:17.35 |
| tor7 | are you booking for me at the same time as well? | 11:17.40 |
| kens | tor7 that was the plan, which is why I haven't done so yet, I wanted to be sure you were OK with it | 11:17.59 |
| | and I'll probably need your passenger details (passport numbner, address etc) | 11:18.21 |
| tor7 | kens: ah, fab. thanks. yes, I'm good with the plan you emailed. | 11:18.40 |
| kens | tor7 OK cna you send me your details by email then (don't want to put that on irc!) and I'll do the booking. | 11:19.12 |
| tor7 | kens: typing up an email now. | 11:20.10 |
| kens | Great, thanks Tor, I'll send you confirmation as soon as I'm done. | 11:20.27 |
| | OK got it, going to book now | 11:22.32 |
| | tor7 ping ? | 11:29.23 |
| | Hmm seems tor has disappeared again | 11:31.45 |
| kens | lunches, will be back shortly. Tor if you get back before me, the web site wants your date of birth too... | 11:35.47 |
| tor7 | kens: sorry, was down in the basement | 11:41.58 |
| kens | OK Tor and I have flights, we'll be arriving on the 6th and departing the evening of the 8th | 12:25.14 |
| tor7 | kens: fab. I'll let miles know to not worry. did you get the seat assignments sorted at last? | 12:28.51 |
| kens | tor7 yes, the web site had not added them to my basket | 12:29.10 |
| | So they were 'pending purchase' | 12:29.18 |
| | You should be abl;e to view them if you use the booking reference | 12:29.35 |
| | On the Delta site of course ;-) | 12:29.52 |
| | I still need to talk to them about the Virgin leg, I just need a short break from wrestling with all this | 12:30.24 |
| | more coffee.... | 12:32.00 |
| Fabio__ | Hi everyone !!! | 13:26.13 |
| | Is there someone involved in MuPDF library development for Android platform ? | 13:26.48 |
| kens | Robin_Watts, paulgardiner you have a customer :-) | 13:27.28 |
| paulgardiner | kens: Ah right. Thanks for the prod. | 13:28.05 |
| | Fabio__: I may be able to help | 13:28.17 |
| Fabio__ | Hi Paul, we just started to use the MuPDF library in order to develop a specific app with PDF functionalities | 13:29.06 |
| | unfortunately we need sticky note support too | 13:29.30 |
| paulgardiner | Does that refer to a type of annotation? | 13:30.19 |
| kens | thinks so yes | 13:30.31 |
| | the post-it note thing I think | 13:30.37 |
| Fabio__ | yes ... exactly we need to add reader comments | 13:31.05 |
| chrisl | Why on earth does Acrobat use different names than the PDF spec - that seems bonkers! | 13:31.22 |
| paulgardiner | chrisl: it is certainly unhelpful. :-) | 13:31.53 |
| Fabio__ | sorry ... the post-it note or other similar functionality that allows a reader to add comments | 13:32.27 |
| Robin_Watts | Fabio__: We (well, paulgardiner really) has been adding support for different annotation types to MuPDF. | 13:33.01 |
| | We support highlight, strikeout, ink, and most recently digital signatures. | 13:33.33 |
| paulgardiner | Because of the difference in naming, I'm still not quite sure which type this is. | 13:33.41 |
| Robin_Watts | We don't support sticky notes yet. | 13:33.56 |
| | Fabio__: Are you a commercial or supported customer ? | 13:34.09 |
| Fabio__ | OK ... we haven't analized the native library yet thererefore at this phase we don't know if post-it notes are supported by native code but the Android Java JNI wrapper | 13:34.21 |
| Robin_Watts | Fabio__: Time for my standard disclaimer then... | 13:34.42 |
| | MuPDF offers a C level API. This is the supported API. | 13:35.00 |
| | We have built various tools/apps on top of this API, including, on android, an example file viewer. | 13:35.25 |
| | As part of that file viewer we have exposed SOME of the api across into java, but just enough for our purposes. | 13:35.47 |
| | The java level is not a supported API. We can't promise you we won't change it at a moments notice. | 13:36.26 |
| Fabio__ | anyway in case Java (+ JNI code) or both Java and C native code must be added we are planning to implement it | 13:36.29 |
| | as we need it | 13:36.35 |
| Robin_Watts | Cool. | 13:36.45 |
| Fabio__ | if you are interested we can align ourselves | 13:37.02 |
| Robin_Watts | I have started to implement a generic java reflection of our C api, but it's not there yet. | 13:37.06 |
| | We'd certainly be interested to see what you come up with. | 13:37.22 |
| | and we can provide pointers if you get stuck, but our priority has to be supported or commercial customers. | 13:37.48 |
| Fabio__ | maybe we can ask help from you as already expert on MuPDF code and inform you about the code we are adding | 13:39.43 |
| Robin_Watts | Fabio__: Just to be clear... you're wanting to build an app that uses MuPDF? | 13:40.26 |
| Fabio__ | in such a way both of us will avoid to implement the same features | 13:40.34 |
| Robin_Watts | Will this be licensed under the GNU GPL ? (i.e. will it be open source etc?) | 13:40.49 |
| Fabio__ | Right ... as I know it should be GPL as the original code | 13:41.34 |
| Robin_Watts | ok, that's fine. As long as you're aware of the license issues. | 13:42.09 |
| Fabio__ | ok ... let me investigate the native MuPDF code ... I have to understand if it already support post-it notes or not | 13:43.09 |
| Robin_Watts | Fabio__: I am pretty sure that it does not. | 13:43.31 |
| paulgardiner | That's right. Not one we currently support, but not a hugely difficult one. | 13:44.50 |
| Fabio__ | OK ... in such a case we must start studying both the MuPDF state-of-the-art and the PDF standard for notes | 13:45.40 |
| Robin_Watts | It's on our list of things to do, but we can't promise when it'll get done. | 13:45.55 |
| Fabio__ | ok ... in such a way as soon as possible I have further questions and news I'll update you | 13:49.28 |
| | thanks a lot for your support | 13:50.03 |
| paulgardiner | tor7: ping | 14:10.49 |
| tor7 | paulgardiner: yes? | 14:11.43 |
| paulgardiner | I had trouble building the iOS app under XCode 4.6. I had to delete the "Generate" target, and use make generate on the command line instead. | 14:14.07 |
| | With the Generate target in place, the MuPDF target failed. | 14:14.28 |
| | Is that unexpected? | 14:14.40 |
| tor7 | paulgardiner: yeah. it ought to just work... :( | 14:16.10 |
| | I rebuilt and jigged things around to work with the latest publicly available Xcode just before the trip | 14:16.39 |
| paulgardiner | It seemed to try to do the generate part as part of the MuPDF build and then complaing about its being armv7 | 14:16.58 |
| tor7 | it should be building the "generate" target as a host build (it's set to use the macosx SDK thingy) | 14:17.26 |
| paulgardiner | Yeah I thought it should | 14:17.40 |
| | Seemed a nice way to do it: make dependent targets | 14:17.59 |
| tor7 | if the generated files don't get built from that step, the makefile will then try to build and run the cmapdump etc tools as arm, which will fail as expected | 14:18.04 |
| paulgardiner | Oh I see | 14:18.18 |
| tor7 | running "make generate" from the command line, using the regular makefile should always work though | 14:18.53 |
| paulgardiner | Ah right. In my case the generated stuff was present because I'd built from the command line, but it still seemed to run | 14:18.55 |
| tor7 | if you do that first, then the Xcode MuPDF target ought to build properly | 14:19.04 |
| paulgardiner | Didn't though | 14:19.13 |
| | I still had to delete the Generate target | 14:19.24 |
| tor7 | *sigh* | 14:19.25 |
| | and if you delete the Generate target, run "make generate" manually, and then build the MuPDF Xcode target, does that work? | 14:19.46 |
| paulgardiner | Yep, that's fine. | 14:19.59 |
| | I'm happy to look into this, but I didn't want to go messing before checking with you | 14:20.38 |
| | Another problem is that the app crashes if I touch the "Library" button - well it move the the library view and then crashes destroying the doc view | 14:22.51 |
| | Is that part of the new locking problem | 14:23.03 |
| | ? | 14:23.05 |
| | Hmmm, I updated XCode very recently. I'm surprised we are different versions. | 14:23.30 |
| tor7 | paulgardiner: do what needs to be done, I thought I'd figured out a way to make it work across versions but apparently not :( | 14:24.17 |
| paulgardiner | Okay. I'll have a go. | 14:25.02 |
| tor7 | might be worth trying to just rip out the threading stuff I've done and recode it single threaded until it works | 14:25.06 |
| | there's a "deprecation" warning on one of the multithreading functions we use | 14:25.40 |
| paulgardiner | tor7: Yeah I wondered that, but I might see if I can trace and understand the problem first - have a little go at least | 14:25.53 |
| | tor7: would it be a bad idea if I split main.m into its various classes? | 14:27.00 |
| tor7 | paulgardiner: no, it's probably time to do so. | 14:27.31 |
| paulgardiner | Oh okay good. | 14:27.43 |
| kens | tor7 I was just looking at transportation at the Maui end. We touch down at 21:44 and the shuttle services stop at 22:00 so we will have to get a taxi | 14:29.19 |
| | Morning Michael | 14:32.10 |
| mvrhel_laptop | good morning kens | 14:32.21 |
| tor7 | kens: okay. it's a good 45 minute drive from the airport. | 14:48.35 |
| kens | cannot drive | 14:48.47 |
| tor7 | even if I could, I wouldn't after a 20+ hour flight | 14:49.24 |
| kens | I'll look at it again nearer the time, there are a couple of options, I quite fancy the idea of being picked up, and its no more expensive | 14:49.46 |
| paulgardiner | robin_watts: Anti-bitmap-churn patch is on paulg/master | 14:50.06 |
| Robin_Watts | kens: The hotel has a shuttle bus? | 14:50.07 |
| kens | I am surprised the taxi frims think that 35 minutes from landing to pickup is enough though | 14:50.13 |
| | Robin_Watts : yes | 14:50.18 |
| | But it stops at 10pm | 14:50.23 |
| Robin_Watts | kens: It's a small airport :) | 14:50.32 |
| kens | Its a resort bus, not a hotel shuttle exactly | 14:50.35 |
| Robin_Watts | ok. | 14:50.48 |
| kens | Its on the westin web page | 14:51.03 |
| Robin_Watts | I'll worry about that nearer the time. | 14:51.13 |
| tor7 | kens: what date are we back in the UK? 10th or 11th? | 14:51.20 |
| kens | I don't think it'll be a preoblem for you Robin_Watts | 14:51.29 |
| | tor7 we arerive on the 10th | 14:51.38 |
| | Depart late on the 8th, arrive early on the 10th | 14:51.52 |
| tor7 | kens: looking at options for my surrounding flight to/from gatwick | 14:52.05 |
| kens | some horrible time like 06:35 or something | 14:52.12 |
| | Yeah I figure you would need to look tor | 14:52.27 |
| | If you really to I think we could get you to Gatwick on the 10th as well, or you could just stay another night here | 14:52.55 |
| Robin_Watts | In postscript: 0 0 moveto 100 0 lineto fill | 14:52.57 |
| | would you expect that to mark the page? | 14:53.04 |
| kens | No | 14:53.06 |
| | the path is not closed | 14:53.16 |
| | But I would check the PLRM for specifics | 14:53.37 |
| Robin_Watts | is that a definite "the page will not be marked", or "it's undefined" ? | 14:53.41 |
| tor7 | kens: doesn't the fill do an implicit closepath? | 14:53.45 |
| kens | tor7 that's why I said I would check the PLRM | 14:54.00 |
| Robin_Watts | tor7: The gs code does an implicit close. But it's a zero width area. | 14:54.26 |
| kens | THere are some odd behaviours for thinkgs like that, and the PLRM is usually pretty good about specifying them, *if* you can find it..... | 14:54.32 |
| | Robin_Watts : yes, but I think its defined as a minimum area,so it may well make a mark if the fill implicitly closes the path | 14:55.01 |
| tor7 | Robin_Watts: yeah. I know that the PDF and XPS rules for what to do with a "0 0 moveto 0 0 lineto fill" differ | 14:55.04 |
| kens | its kind of like a 0 setlinewidth stroke | 14:55.14 |
| | let me get my PLRM | 14:55.59 |
| paulgardiner | tor7: Oh yeah, another thing I meant to ask: does Artifex have an iOS developer account? | 14:56.22 |
| kens | OK hte PLR does clearly say that open paths are implicitly closed | 14:56.43 |
| tor7 | kens: okay, round trip CPH-LGW 5th - 11th is about GBP70 - GBP90 depending on time of day | 14:56.57 |
| kens | THat's not bad | 14:57.06 |
| tor7 | kens: I'd say that's ridiculously cheap. flying out the 4th I can get it as low as GBP60 | 14:57.31 |
| | it costs more to go to stockholm... | 14:57.42 |
| kens | Yeah that's *really* cheap | 14:57.46 |
| henrys | finally the last day | 14:58.31 |
| kens | :-)# | 14:58.37 |
| henrys | and go home to a biblical flood | 14:58.50 |
| Robin_Watts | henrys: You know you love it really. | 14:58.52 |
| | A shape is scan-converted by painting any pixel whose square region intersects | 14:59.11 |
| | the shape, no matter how small the intersection is. This ensures that no shape | 14:59.12 |
| | ever disappears as a result of unfavorable placement relative to the device pixel | 14:59.14 |
| | grid, as might happen with other possible scan conversion rules. The area covered | 14:59.16 |
| | by painted pixels is always at least as large as the area of the original shape. This | 14:59.17 |
| | rule applies both to fill operations and to strokes with nonzero width. Zero-width | 14:59.19 |
| | strokes are done in a device-dependent manner that may include fewer pixels | 14:59.21 |
| kens | Robin_Watts : the PLRM says that any pixel whose region intersects the shape is painted. so your fill should paint a 1 pixel wide line | 14:59.21 |
| Robin_Watts | than the rule specifies. | 14:59.22 |
| | kens: ^ From the PLRM. | 14:59.23 |
| | Right. | 14:59.32 |
| kens | Obviously we were reading the same apragraph :-) | 14:59.42 |
| Robin_Watts | And at 600dpi, the gs scan converter chooses not to mark the page at all. | 15:00.12 |
| | at 300dpi, it does mark it for this example. | 15:00.23 |
| paulgardiner | If you were doing antialiasing then the intensity with which to paint might be zero | 15:00.27 |
| kens | OK well that is 'wrong' though the sepc does say that this is not part of the language. | 15:00.37 |
| mvrhel_laptop | bbiab | 15:01.15 |
| Robin_Watts | paulgardiner: yeah. | 15:01.17 |
| kens | Nevertheless we do try to mimic the Adobe output and it seems to me it should paint the line | 15:01.19 |
| Robin_Watts | kens: yes, but to make it do so is... hard. | 15:01.46 |
| kens | I'll have to take your word for that, not an area I've looked at in GS | 15:02.12 |
| Robin_Watts | Horizontal lines don't get represented in the edge list of the scan converter. | 15:02.30 |
| kens | :-( All I can really say is that it looks like it should. Although its a really dumb way to construct a file | 15:03.54 |
| | I see its made by Acrobat Distiller, so its probably generated that way in the CAD package | 15:05.01 |
| Robin_Watts | I wonder if I can spot horizontal lines in the fill adjust !=0 case, and put two small vertical lines into the edge list. | 15:05.26 |
| | For each edge of the fill adjusted area. | 15:05.45 |
| kens | Don't know, it kind of sounds like it might have side-effects | 15:05.52 |
| Robin_Watts | I'd have to be careful with the windings. | 15:06.13 |
| | I could add code to spot this particular case. (A path that is purely a horizontal line, with non-zero fill adjust) | 15:08.49 |
| | That would be safe to fix. | 15:08.59 |
| kens | And vertical are OK because they are in the edge list, waht about diagonals ? SOunds like they should hit the edge list too ? | 15:09.39 |
| | In which case spotting degenerate horizontal lines sounds like a good special case. | 15:10.10 |
| Robin_Watts | kens: yeah. it's just horizontals that are a problem. | 15:10.57 |
| | But I've run into this before and backed out of it, with cases like: | 15:11.10 |
| | 0 0 moveto 100 0 lineto 100 100 lineto 100 0 lineto closepath | 15:11.30 |
| | That's a zero area shape. | 15:11.40 |
| kens | Ah because of the closing segment ? | 15:11.43 |
| | Oh no I see | 15:11.55 |
| Robin_Watts | the vertical section shows up, the horizontal section doesn't. | 15:12.12 |
| kens | On GS or Adobe ? | 15:12.29 |
| | looks to me like both lines should mark | 15:13.09 |
| Robin_Watts | on adobe, I believe both mark. | 15:13.28 |
| | on gs, I'm not so sure. | 15:13.33 |
| | certainly I've seen differences in clippaths defined like that. | 15:13.50 |
| kens | Well, it looks to me like both should, if we aren't then I think we're wrong (for particular values of wrong) | 15:13.53 |
| | clippaths would be different, unless they also do an implicit close | 15:14.17 |
| | Acrobat does show 2 lines for your test case there (provided you first turn off smoothing....) | 15:17.46 |
| | If I set the path to be the clipping paht with 'clip' and then fill the area, I just get the two lines drawn again | 15:20.38 |
| | Your original case, which doesn't do a closepath draws a horizontal line, so it seems like 'clip' does an implicit closepath as well. | 15:22.31 |
| | Oh yes, the clip operator details even say that. | 15:22.54 |
| paulgardiner | tor7: More wierdness - tried a completely fresh checkout and then just running the Generate target. cmapdump segfaults. Tried rerunning the command under gdb and I get "The target crashed on startup, maybe the shell is crashing" | 15:28.36 |
| Robin_Watts | kens: Actually, horizontal lines *do* get put into the scan converter active edge list, but only if fill adjust is being used (makes sense), and if they are positioned such that the fill adjust makes them cover a pixel centre. | 15:32.35 |
| kens | Hmm, well that isn't what the PLRM says.... | 15:32.52 |
| Robin_Watts | And the lines in this particular file, at 600dpi, are positioned EXACTLY so that they don't quite seem to cover any centres. | 15:33.12 |
| | The math is going wrong somehow. | 15:33.22 |
| kens | Typical.... | 15:33.23 |
| Robin_Watts | adjust_below is 127, adjust_above = 128 | 15:34.08 |
| paulgardiner | tor7: I guess our makefile, when run from the project, creates a malformed exectuable, although it works when run from the command line | 15:34.35 |
| tor7 | paulgardiner: wow. that's bad. | 15:41.15 |
| | paulgardiner: I know Xcode sets a number of magic environment variables when you run "make" from inside it. | 15:41.37 |
| | it might be making an 386 executable that can only be run from inside the iphone simulator... | 15:42.22 |
| paulgardiner | Yeah. I've added -d to see if I can get it to give me some indication what is going wrong | 15:42.35 |
| tor7 | if the setting to build it as a macosx target isn't working | 15:42.42 |
| | paulgardiner: change the argument in the Generate target to run "make verbose=yes" | 15:43.03 |
| paulgardiner | Right ta | 15:43.13 |
| tor7 | and as a last ditch attempt, uncheck the "set environment variables" option | 15:43.25 |
| paulgardiner | Unchecking that seems to help. Would have thought that make was able to pick up all it needed fromt the environment in any case | 15:54.24 |
| kens | chrisl I may punt #694591 to you, I suspect its someone using system sahred libraries on Linux and if they use our code it will probably work. | 16:07.04 |
| Tomek_ | kens, chrisl: as a follow-up to my earlier question - what you suggested with mudraw works for me and I will not need to try statically linking with X11; thank you very much for advice and help! | 16:41.57 |
| kens | You're very welcome Tomek_ | 16:42.12 |
| | thansk for letting us know | 16:42.20 |
| | Goodnight all | 16:49.49 |
| Robin_Watts | henrys: ping | 17:00.51 |
| | I thought pxl always ran with fill_adjust = 0 ? | 17:01.05 |
| | oh., no, it's a horrid hack for pcl, right? If the resolution is low, fill_adjust is +/- 1/3 or something ? | 17:01.38 |
| sicos | hi | 17:22.06 |
| | me again with a question. I want to see if a pdf has more then 2 color (black and white). I know that I can count the colors on a pdf with the inkcov device. But is it also possible to count the colors when I'm processing a pdf with a run command from postscript? | 17:23.22 |
| Robin_Watts | sicos: mvrhel_laptop has just been working on a mechanism for determining if a page is 'neutral' colored or not. | 17:24.43 |
| | So with that you can run a page and find out whether it's black/white/greyscale or whether it contains colors. | 17:25.13 |
| | Is that what you're after ? | 17:25.19 |
| russcriptor_ | hey guys it seems make defaults to build=debug | 17:29.44 |
| | and even if I do make build=release | 17:30.14 |
| | I can't seem to find the corresponding install command | 17:30.26 |
| Robin_Watts | make build=release install ? | 17:30.49 |
| russcriptor_ | Robin_Watts: that seems to work thanks | 17:31.17 |
| | Robin_Watts: what would be the difference between the debug and release builds? | 17:31.32 |
| | Robin_Watts: and native | 17:32.12 |
| Robin_Watts | debug builds have debugging information built in. | 17:33.28 |
| | what's 'native' ? | 17:33.36 |
| russcriptor_ | make build=native | 17:33.58 |
| | although that one doesn't actually work | 17:34.04 |
| | so a release build would theoretically be faster? | 17:34.25 |
| sicos | Robin.. sorry for the late answer.. was making some coffee. | 17:35.36 |
| | Yes that is what I'm after | 17:35.42 |
| Robin_Watts | russcriptor: yes | 17:35.45 |
| | sicos: So you need to talk to mvrhel_laptop. Michael, you about? | 17:36.45 |
| sicos | Because of a stupid problem that they have at the company where I'm working I did make a PDF (with text and images) to a PDF with only images converter | 17:37.03 |
| | I did that with ghostscript | 17:37.09 |
| | first I load a pdf and convert it to png16m | 17:37.19 |
| | then I convert the png images to pdf again with pdflib | 17:37.29 |
| | I output the png images with 300 dpi | 17:37.41 |
| | When I have a almost neutral page.. with only text (like an E-mail) and I convert that to png color and back again to a pdf the quality degrades some | 17:38.39 |
| | When I use the pnggray device the quality is better | 17:38.52 |
| | So that is why I want to detect if a pdf has colors or not | 17:39.06 |
| Robin_Watts | I can see no reason why that should be the case. | 17:39.11 |
| sicos | When it has color (mainly because of images) I use png16m and when it doesn't have colors i want to use pnggray | 17:39.31 |
| Robin_Watts | I understand that it's more efficient to produce greyscale output when you have greyscale input. | 17:40.11 |
| | but I can see no reason why there should be a quality difference. | 17:40.22 |
| sicos | it's also a lott faster | 17:40.22 |
| | since converting a pdf with 20 pages to png16m takes some time | 17:40.48 |
| Robin_Watts | yes, I can imagine there is a speed difference too, but I'd like to see an example of your asserted quality loss. | 17:41.39 |
| sicos | I know that an image from a page wont be as good as the original pdf but that is what you get with images... no antialiasing.. etc.. | 17:42.04 |
| russcriptor_ | mudraw | 17:42.48 |
| | we've just switched to that and it's working much better then ghostscript | 17:43.05 |
| sicos | I use -r300 as quality | 17:43.13 |
| Robin_Watts | sicos: You claimed that a greyscale PDF converted to a greyscale image had better quality than a greyscale PDF converted to a color image. I'd like to see evidence of that. | 17:43.17 |
| sicos | that should be good enough? | 17:43.20 |
| | I'm at home at the moment.... and just started my first day of my holiday... so I can't send you an example for the next 2 weeks | 17:44.00 |
| Robin_Watts | If you were to use mudraw for the PDF -> PNG stage, you'd get antialiasing too. | 17:44.06 |
| | sicos: Fair enough. have fun. | 17:44.21 |
| sicos | Robin: Is it also possible to control mudraw from code? | 17:45.16 |
| russcriptor_ | Robin_Watts: I want to write a C function that does the conversion | 17:45.23 |
| | yes | 17:45.26 |
| | I have teh same question | 17:45.32 |
| | example.c has a decent example | 17:45.39 |
| | but it's pretty limited | 17:45.46 |
| | I want to be able to pass in resolution and max width | 17:46.04 |
| Robin_Watts | sicos: You can call mudraw as a shell command - presumably that's what you're doing to call gs? | 17:46.10 |
| sicos | Don't laugh but I still use VB6 to control the gs32.dll through it's api.. it should be ported to C# some day but it's an lott of code | 17:46.33 |
| Robin_Watts | russcriptor_: Well, that's just a question of tweaking example.c :) | 17:46.33 |
| | sicos: Ah, well, mupdf does not have a C# interface. Or at least, not one that I know of. | 17:46.58 |
| sicos | And it is higly optimized for speed...in 2006 c# was terrible when you talked with a lott of com objects | 17:47.04 |
| Robin_Watts | and there is 0% chance of me doing one :) | 17:47.14 |
| sicos | So I did make everything in VB6... rock stable and fast | 17:47.37 |
| | Robin: youre on linux ? | 17:48.37 |
| Robin_Watts | sicos: Nope. | 17:49.00 |
| sicos | Windows 8 on this side... but I haaaaaaaaaaatttttttttteeeeeeeeeeeeee the new start menu.. thank god that they did make classic shell | 17:49.16 |
| | I mean classic start menu | 17:49.33 |
| Robin_Watts | I'm on Win 7. I won't be upgrading any time soon. | 17:49.41 |
| sicos | It's also not worth it | 17:50.12 |
| Robin_Watts | I have Win 8 in a vmware image. | 17:50.29 |
| sicos | The only advantage I find usefull with Windows 8 is it startup speed | 17:50.32 |
| | But to get back to the subject... mudraw will make better color png's then ghostscript? | 17:51.49 |
| Robin_Watts | mudraw antialiases in a way that gs doesn't. | 17:52.30 |
| | (gs can be told to antialias, but I think the results are generally better for screen output from mupdf) | 17:52.55 |
| sicos | I also tried tiffscaled24 but that wasn't any better... and the manual said it also used antialiasing | 17:53.32 |
| | and the output files were huge | 17:53.44 |
| Robin_Watts | tiffscaled24 works by rendering larger and scaling down. | 17:54.32 |
| | tif files are uncompressed. | 17:54.52 |
| | but you'd expect the PNG files to be larger if antialiasing was used, as there are more colors to represent. | 17:55.14 |
| sicos | I'm going to have a quick look into mupdf | 17:55.19 |
| | I don't mind if it gets some bigger... in my case the quality counts | 17:55.50 |
| | but there is a diference between quality and huge files | 17:56.07 |
| henrys | Robin_Watts: here now⦠what's up | 17:59.28 |
| Robin_Watts | henrys: pcl does use fill_adjust, right? | 18:02.25 |
| sicos | hum | 18:13.32 |
| | I did a test with mudraw and a png is much better then when i do the same with ghostscript | 18:13.56 |
| | why is there so much difference? | 18:14.06 |
| | is the png engine in mudraw so much better? | 18:14.39 |
| Robin_Watts | The rendering engines in mupdf and ghostscript are completely different. | 18:18.34 |
| | mupdf is targetted at screen output, whereas gs is targetted at print devices. | 18:19.08 |
| | hence they have different tradeoffs. | 18:19.15 |
| | If you give the appropriate flags to gs you can force to it antialias too (-dGraphicsAlphaBits=4 -dTextAlphaBits=4) | 18:19.58 |
| russcriptor_ | sorry guys was away | 18:21.14 |
| | we've switched to mudraw and the quality and speed are much better | 18:21.27 |
| | hoping to get even better speed by calling the c methods directly | 18:21.45 |
| | so back to my question | 18:22.16 |
| | I was able to integrate with example.c and it seems pretty straightforward, although its doing some things I don't need such as rotation and not doing things I need such as resolution and maxwidth | 18:23.14 |
| | is there a better example somewhere? | 18:23.21 |
| | I treid looking at mudraw.c but it seems pretty complex | 18:23.34 |
| henrys | Robin_Watts: xl does and pcl doesn't | 18:24.51 |
| | parts of my town are being evacuated ⦠not good | 18:25.30 |
| russcriptor_ | henrys: you in CO? | 18:25.46 |
| | Robin_Watts: is there documentation somewhere for the mupdf api? | 18:30.24 |
| | http://www.mupdf.com/docs/overview seems pretty limited | 18:30.42 |
| sicos | lost my internet connection | 18:34.17 |
| | Why is there so much difference between mudraw pdf and ghostscript? | 18:34.32 |
| | i mean png | 18:34.37 |
| | When I export a page to png with mudraw it is almost exactly as the original one in the pdf | 18:35.13 |
| | when I do it with ghostscript its more blurry and the fonts look a little bit different | 18:35.47 |
| | d:\>gswin32c -q -dQuiet -dBATCH -dNOPAUSE -sDEVICE=png16m -r300 -dAlignToPixels= 0 -dTextAlphaBits=4 -dGraphicsAlphaBits=4 -sOutputFile=d:\testt%d.png d:\test.pdf | 18:36.10 |
| | I come a lott closer when i turn on the alphabits | 18:36.23 |
| | but still not as close as mudraw | 18:36.29 |
| Robin_Watts | russcriptor_: The documentation is pretty much in the form of example files. | 18:36.36 |
| | mudraw.c shows extensively how to call the API. | 18:36.48 |
| | sicos: read the logs for my reply to you earlier. | 18:37.11 |
| sicos | sorry no log... i use freenode web irc | 18:37.43 |
| Robin_Watts | http://ghostscript.com/irclogs/current.html | 18:38.44 |
| | henrys: Crumbs. | 18:39.48 |
| | Is your house OK? ISTR you saying it was on a hill? | 18:40.00 |
| sicos | Robin: sorry but I don't know what to find in the logs | 18:41.15 |
| Robin_Watts | sicos: You asked your question at 18:14. I replied at 18:18. | 18:42.01 |
| | You then asked again at 18:34, and rather than reply again, I thought it would be faster to point you to the logs for my original answer. I guess not. | 18:42.44 |
| sicos | ah ok.. sorry I missed that one... | 18:43.07 |
| | I tried the -dGraphicsAlphaBits=4 -dTextAlphaBits=4 commands.. it makes it a lott better... but mudraw is still better so I probably going to stick with that one | 18:44.08 |
| | Do the -dGraphicsAlphaBits=4 -dTextAlphaBits=4 commands also work on the jpeg device? | 18:47.40 |
| Robin_Watts | sicos: Yes. | 18:48.58 |
| | but why in gods name would you render stuff with text in to the jpeg device? :) | 18:49.47 |
| sicos | Like I told before... because of stupid software package called streamserve | 18:50.47 |
| | It is a print system to make letters and things like that | 18:51.10 |
| | When you feed it a pdf with text it tries to convert the pdf to its internal document type | 18:51.33 |
| | as soon as it finds a font that it doesn't know it produces an error... and thats it | 18:51.52 |
| | no more processing | 18:51.57 |
| | So they asked me to solve it by converting a pdf to a pdf with only images | 18:52.11 |
| Robin_Watts | sicos: sure, but you have a fix that works with PNGs, right? | 18:52.11 |
| | Using JPEG for anything with text in is HUGELY bad. | 18:52.27 |
| sicos | Yeah thats right | 18:52.32 |
| | But the conversion system that i did make also produces tiff's and sometimes jpg that they use to ocr scanned pdf's | 18:53.01 |
| | so thats why I asked about the -dGraphicsAlphaBits=4 -dTextAlphaBits=4 commands | 18:53.13 |
| Robin_Watts | sicos: Ah. Well, mudraw will read jpegs and tiffs :) | 18:53.44 |
| sicos | well in this case they need tiff's as output | 18:54.56 |
| | The conversion server that I did make started as a very simple thing but grew a lott in 5 years. from just converting html to pdf (with wkhtmltopdf) to a system that accepts about 50 types of files to pdf, pdfa1/b pdfa2/b tiff ccitt4, tiff grayscale lzw, jpg and png | 18:56.19 |
| | and the latest add on is the pdf to pdf with images conversion | 18:56.31 |
| | it also concatenates pdf's, splits them, makes pdf's for e.g. with a selective range of pages from a source pdf | 18:57.38 |
| | it also can add attachments into a pdf... e.g. convert a excel document to pdf and then also attach the excel file into the pdf | 18:58.08 |
| | For all the pdf things i used ghostscript... did write postscript code to solve most of the things | 18:58.35 |
| | it's even the way i started learning postscript | 18:58.58 |
| | Well I'm going to watch some tv... already 9pm overhere in the Netherlands... I guess all of you are from the US? | 19:01.29 |
| | bye and thanks for the chat | 19:04.08 |
| mvrhel_laptop | bbiab | 19:33.08 |
| marcosw | 2 | 23:57.17 |
| | Forward 1 day (to 2013/09/13)>>> | |
IRC Logs