| | <<<Back 1 day (to 2013/05/14) | 2013/05/15 |
| kens | chrisl I was looking at using the bbox device to make an eps2write device. THe code in the vector device doersn't work. The reason is that when a device is opened, there is no way to push another device ahead of it in the pipeline. | 08:56.59 |
| | So, to make an eps2write device, I would need to make a 'dummy' device that just forwards everything. That creates a ps2write device and a bbox device, and forwards the bbox device to the ps2write device. It then forwards everything itself to the bbox device. | 08:57.59 |
| | Kind of icky, but I think it will work. | 08:58.25 |
| Robin_Watts | tor8: Morning. | 09:08.49 |
| | http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=ddb4dd4ab1ddeb25f829caa0d3dfc208c7eaf68e | 09:12.51 |
| paulgardiner | Robin_Watts, tor8: I have some code to do with signatures that I want to commit, but I'm not sure what to do about the reliance on openssl. | 09:43.36 |
| | Do we treat it like the other main thirdparty libraries, or special case it like v8? | 09:44.19 |
| | For now I could protect it with defines, but then it doesn't get cluster tested. | 09:44.40 |
| tor8 | how big is openssl and how difficult is it to build? | 09:45.15 |
| Robin_Watts | what's the license on openssl, and how separable is it? | 09:46.46 |
| tor8 | if it's not too horrible, we could make it a regular thirdparty dependency and rip out our own copies of crypto stuff | 09:47.31 |
| paulgardiner | My checked out and built version is 227MB, but I guess you want freshly checked out size | 09:48.10 |
| tor8 | the source archive is 4MB | 09:48.30 |
| paulgardiner | Building doesn't seem to have been a problem. Uses make on linux, and nmake on windows. | 09:48.38 |
| tor8 | and the compiled libcrypto.a is 4.2MB!!! | 09:49.04 |
| | eek. | 09:49.06 |
| | but I suspect we'll only actually need a fraction of that in reality | 09:49.29 |
| paulgardiner | tor8: I've made only the apps rely on it, not the library if that helps. | 09:49.50 |
| tor8 | 2.3MB when stripped | 09:49.53 |
| paulgardiner | It's released under two lincenses, both BSD-style | 09:50.16 |
| tor8 | paulgardiner: if we decide to ship with it, we may as well make the library depend on it as well. | 09:50.18 |
| paulgardiner | tor8: that would be handy. Would allow me to roll some of my app code into the library for reuse | 09:50.50 |
| tor8 | not saying to make the library depend on openssl just for the sake of it, but if it simplifies matters and removes the need for our own copies of crypto functions | 09:51.06 |
| | paulgardiner: ideally it should be possible to build the thirdparty libs with our own makefiles (see Makethird) | 09:51.41 |
| paulgardiner | It would allow me to add a "check signature" function to the library. | 09:51.47 |
| tor8 | third party configure scripts often do terrible things like build shared libraries etc | 09:51.53 |
| paulgardiner | tor8: I'd imagine setting up our own make file wouldn't be hard. | 09:52.23 |
| | And would help us pick out only what we need. | 09:52.35 |
| tor8 | paulgardiner: I'll see about doing that today then, gives me another reason to not have to touch Gtk+ for one more day | 09:52.48 |
| paulgardiner | tor8: oh okay, but I'm happy to have a go if you're at all worried about the withdrawal symptoms you mgiht experience when away from Gtk+ :-) | 09:54.23 |
| tor8 | paulgardiner: I need to set up a local git mirror of the upstream openssl git repo first and hook all that in too | 09:56.46 |
| paulgardiner | Ah I see | 09:57.07 |
| tor8 | it's easier for us to just use an upstream git if it exists, but we want to host our own in case external servers go down or move | 09:57.31 |
| | ok, there should be a git.ghostscript.com/thirdparty/openssl.git mirror live now | 10:00.42 |
| paulgardiner | tor8, Robin_Watts: Another thing I've been meaning to mention: I'm getting increasingly disillusioned with the fz_interactive interface. It is looking more pdf specific, the more I add to it, and it seems a shame when performing pdf-specific tasks not to be able to return arrays as pdf arrays and generally make use of all the nice pdf object stuff. | 10:01.14 |
| tor8 | paulgardiner: zap it! I already gave you my blessing :) | 10:01.45 |
| Robin_Watts | go for it. | 10:01.53 |
| paulgardiner | Great. I may just check with henrys about taking the time to do it. Could take a day, with all the fiddling. | 10:02.37 |
| Robin_Watts | tor8 The link i gave earlier for review... my bmpcmp area now contains the diffs it causes - you can see they are all progressions. | 10:14.47 |
| tor8 | Robin_Watts: pdf_copy_gstate looks odd to me, you only drop the stroke state but clobber all fields | 10:26.51 |
| Robin_Watts | tor8: yes, that does look odd, doesn't it. | 11:42.50 |
| | testing a fix now. | 12:00.27 |
| | tor8: http://git.ghostscript.com/?p=user/robin/mupdf.git;a=commitdiff;h=bd30dda062582e6172d5c1db8fe1470d644c1d05 | 12:18.06 |
| | Same diffs as before. | 12:18.13 |
| tor8 | Robin_Watts: feel free to push. | 12:55.55 |
| Robin_Watts | Thanks | 13:04.58 |
| | tor8: (For the logs) and paulgardiner (and sebras if interested): I've taken mvrhel's winrt branch and squashed it to a single commit. | 13:36.58 |
| | It's on robin/winRT if you're interested. | 13:37.07 |
| paulgardiner | Ah right. Handy. | 13:37.25 |
| Robin_Watts | I've also removed a spurious change to the version of the win32 solutions and added a missing newline. | 13:37.28 |
| | and rebased it up to date. | 13:37.35 |
| | I'm looking through it now, and realising how odd C# is. | 13:37.54 |
| | Oh, I may need to tabify it too. | 13:39.00 |
| paulgardiner | I can't at the moment see anything to distinguish it from C++ | 13:49.20 |
| Robin_Watts | well, I count C++ as being pretty odd :) | 13:52.35 |
| paulgardiner | Very true | 13:52.48 |
| Robin_Watts | I dislike this "#pragma once" thing. MS introducing boilerplate for something which can be solved perfectly without it. | 14:10.30 |
| | but then I dislike the precompiled header stuff too. It's extra voodoo complexity that doesn't save us anything. | 14:11.36 |
| | tor8; see the logs | 14:14.59 |
| | DocumentPage::DocumentPage sets: this->{Image,Height,Width,Zoom,Content} | 14:19.45 |
| | DocumentPage.h defines the DocumentPage class with "image,height,width,zoom,content" (lower case). | 14:20.11 |
| kens | Hmm marcos adsded a new cluster mode 'singlepagePDF' | 14:20.32 |
| Robin_Watts | Oh, it's getters and setters. | 14:20.45 |
| tor8 | Robin_Watts: the c++ reference counted ^ decorator is pretty weird too | 14:20.54 |
| Robin_Watts | fixes indentation, and adds some pretty blank lines for tor8. | 14:25.48 |
| | I am disquieted by the "Block on the async call" loops. | 14:26.34 |
| | At the very least they should sleep. | 14:26.44 |
| kens | marcosw ping | 14:27.38 |
| Robin_Watts | WTF? | 15:00.08 |
| | The mutex locking in the winrt stuff looks way wrong. | 15:00.25 |
| sebras | Robin_Watts: what terrible failure? | 15:00.30 |
| Robin_Watts | It looks like mvrhel has a 'critical section' rather than a mutex, and taking any lock is implemented by entering that critical section. | 15:01.11 |
| | I suspect that a criticalsection can enter itself without any problem, and will nest nicely, so he won't get broken behaviour, but it'll cause lots more cross thread locking than he needs. | 15:02.01 |
| chrisl | I thought that was the normal locking method on Windows - IIRC, it's how pthreads on Windows works | 15:02.54 |
| Robin_Watts | chrisl: The MuPDF locking method is void lock(void *user, int lock) where user = the apps supplied opaque pointer, and lock = which lock to take. | 15:03.34 |
| | mvrhel ignores lock. | 15:03.42 |
| | hence all 4 of our locks are mapped onto 1. | 15:04.11 |
| chrisl | Ah, that's not optimal | 15:04.58 |
| Robin_Watts | and his use of mu_cookie seems 'odd'. | 15:07.18 |
| henrys | paulgardiner: fine by me, it seems like frequent breaks from the signature business is a good thing ;-) | 15:15.11 |
| paulgardiner | henrys: that is so true :-) | 15:15.59 |
| rafcloesen | Hi, regarding MuPDF, I have gone through the manual and googled a bit. I couldn't find a way to view the index of a pdf in mupdf. Is this possible? | 15:35.17 |
| Robin_Watts | rafcloesen: You mean the "outline" of a PDF ? | 15:35.49 |
| rafcloesen | I am using 0.9-2 on Debian | 15:36.03 |
| Robin_Watts | rafcloesen: Ah, right, you're using the linux viewer. | 15:36.24 |
| | No, the linux viewer doesn't have an outline view mode | 15:36.36 |
| | MuPDF is a set of portable C libs that have all the brains in them. | 15:36.52 |
| rafcloesen | I guess, I have a book in pdf, and need the links to the chapters and sections that some pdf viewers show on a sidebar | 15:36.52 |
| Robin_Watts | Plus we provide various tools that are thin wrappers around this lib. The linux viewer is one such tool. | 15:37.18 |
| | The core libs are capable of extracting/displaying the outlines from a PDF - indeed, we use that in the Android and iOS viewers. | 15:37.51 |
| | The current linux viewer doesn't have the required code in though. | 15:38.01 |
| | hey mvrhel_laptop. I'm just writing you an email. Where are you at the moment? | 15:38.24 |
| mvrhel_laptop | in the apple building | 15:38.47 |
| | Robin_Watts: ^^ | 15:38.56 |
| Robin_Watts | oh, you made it there already. | 15:39.06 |
| marcosw_ | the meeting starts in 20 minutes | 15:39.19 |
| mvrhel_laptop | I see that you have found issues in my code. I figured I did not do something correct | 15:39.59 |
| Robin_Watts | mvrhel_laptop: On the whole it looks great. | 15:40.08 |
| mvrhel_laptop | but that is me reading the documentation and trying to see what needs to be done based on that | 15:40.11 |
| Robin_Watts | (In as much as it's C++ and my comprehension of such is challenged :) ) | 15:40.27 |
| | The things I've spotted that are wrong are minor. | 15:40.48 |
| | easily fixed. | 15:40.57 |
| rafcloesen | Robin_Watts: Thank you, I will check out the API then. | 15:41.16 |
| Robin_Watts | rafcloesen: We are working on a new linux viewer, based on GTk, but no promises when it will be ready. | 15:41.41 |
| mvrhel_laptop | Robin_Watts: please feel free to fix the interface stuff to mupdf | 15:43.54 |
| rafcloesen | Robin_Watts: How can I find out when it is done? is there a mailing list or newsgroup? | 15:43.54 |
| Robin_Watts | mvrhel_laptop: I can certainly have a look at the locking. | 15:44.37 |
| | I think I need to talk to you about the cookie to check I'm not missing something cunning. | 15:44.54 |
| mvrhel_laptop | I have one update here that I need to get it which is the progress bar stuff. | 15:46.16 |
| | the cookie stuff is not used really yet | 15:46.23 |
| | so there is nothing cunning going on there | 15:46.47 |
| Robin_Watts | mvrhel_laptop: Right. I wasn't ever imagining that we'd need to allocate the cookie. It can just be a stack based thing, I think. | 15:46.55 |
| mvrhel_laptop | ok | 15:47.04 |
| | I am glad that I didn't have you look at the code about a month ago.... ;) | 15:47.28 |
| | you really would have flipped | 15:47.41 |
| Robin_Watts | but (aside from the size of the xaml and the project files etc), it's impressively small. | 15:47.52 |
| | (not saying "you haven't done much", saying "it's worked out quite neatly and compactly") | 15:48.29 |
| mvrhel_laptop | yes. once I figured out how to leverage a lot of the windows objects and methods on the viewer side things came together nicely | 15:48.38 |
| Robin_Watts | I've got to pop out for a bit. Will look at the locking when I get back. | 15:49.36 |
| | Have fun at the talk! | 15:49.41 |
| mvrhel_laptop | ok thanks | 15:49.43 |
| henrys | catches up on the summit slides given so far. | 15:51.54 |
| kens | OK off now goodnight all | 16:27.12 |
| vtorri_ | hello | 16:51.50 |
| ghostbot | niihau, vtorri_ | 16:51.50 |
| vtorri_ | good bot | 16:51.54 |
| | about mupdf, pdf_load_page() loads a page by passing the number of the page | 16:52.56 |
| | does mupdf store it ? or must i store it ? | 16:53.18 |
| henrys | marcosw, marcosw_: are you able to irc from the meeting? | 17:21.15 |
| marcosw_ | only using a web irc client (which doesn't alert me when you type my handle) | 17:22.05 |
| mvrhel_laptop | he had to be alerted by me | 17:22.17 |
| marcosw_ | luckily mvrhel_laptop is sitting next to me, so he didn't even have to get up to poke me. | 17:23.08 |
| henrys | marcosw_:I'm worried about alerting us here if you guys need help. I don't know if Robin_Watts or for8 will be around that late if a strange mupdf thing comes up but a few of us will be here. | 17:24.20 |
| | mvrhel_laptop: ^^^ | 17:24.46 |
| marcosw_ | henrys: we will just make stuff if someone asks us a question we don't know the answer to. | 17:25.28 |
| henrys | marcosw_:good plan | 17:26.17 |
| marcosw_ | That's something you learn now do to do at graduate school :-) | 17:26.20 |
| | ^now^how | 17:26.25 |
| henrys | marcosw_:hence Doctor of "Philosophy" | 17:27.40 |
| mvrhel_laptop | yes. we can both pontificate about stuff we dont know about | 17:29.06 |
| tor8 | vtorri_: the page or the page number? | 17:29.53 |
| marcosw_ | henrys: you are missing a great discussion about clouds | 17:30.01 |
| vtorri_ | tor8, the page number | 17:30.08 |
| henrys | marcosw_:I did read the slides | 17:30.18 |
| tor8 | vtorri_: then I'm not sure I understand what you mean | 17:30.43 |
| marcosw_ | we are stuck on slide 9. We had made it to slide 11 but had to go back. | 17:30.54 |
| tor8 | you can't get the page number from the fz_page struct if that's your question | 17:31.20 |
| vtorri_ | tor8, i call pdf_load_page(my_page, 8); Is there a mupdf call that would return 8 ? | 17:32.00 |
| tor8 | vtorri_: no. | 17:32.18 |
| vtorri_ | ok, thanks | 17:32.23 |
| henrys | marcosw_:I didn't quite understand the printing to imaging change there and it seems that is very important so I'm hopelessly lost. | 17:32.34 |
| | marcosw_, mvrhel_laptop how does tkamppeter give his talks? | 17:36.05 |
| mvrhel_laptop | henrys: Apparently over the phone He has not done too much today | 17:36.40 |
| henrys | mvrhel_laptop, marcosw_ :once I see the semantic model in UML I know the ship is likely to sink. | 17:42.48 |
| tkamppeter | henrys, marcosw_, mvrhel_laptop: If you want to give a talk on the Summit via phone, e-mail the slides as PDF to Mike Swwet and call in at least some minutes before your talk is scheduled. Also connect to WebEx but note that a browser under Windows/Mac/Linux needs a Java plug-in, under Android you need to install the App WebEx Meetings from Cisco and use the meeting number 626 868 486 and password pwg123. | 17:50.01 |
| | henrys, marcosw_ mvrhel_laptop, Mike will project the slides in the room with a Mac and let the Mac share its screen through WebEx, so you will see the slides and simply talk on the phone and say "Next" to get to the next slide. | 17:51.44 |
| | henrys, marcosw_ mvrhel_laptop: No one of you is on the OpenPrinting Summit in person? | 17:52.14 |
| mvrhel_laptop | tkamppeter: I am here and so is marcosw | 17:52.28 |
| | I could give it over the phone, but I am here | 17:52.42 |
| tkamppeter | mvrhel_laptop, OK, much better. | 17:54.08 |
| marcosw_ | I'm here as well but I'm viewing the slides by connecting via vnc to my home computer that is running WebEx :-) | 17:54.24 |
| henrys | marcosw_:I thought you were there | 17:56.12 |
| | aren't you in the room with the presentation? | 17:56.49 |
| Robin_Watts | mvrhel_laptop: If a complex mupdf question comes up, please feel free to call me. | 17:57.11 |
| mvrhel_laptop | he is right here beside me | 17:57.16 |
| Robin_Watts | "Phone a friend" :) | 17:57.18 |
| mvrhel_laptop | Robin_Watts: yes like that show | 17:57.22 |
| marcosw_ | I'm physically at Apple. | 17:58.11 |
| tkamppeter | marcosw_, why do you watch the slides that awkward way? Must be really slow. | 18:04.34 |
| | henrys, what do you mean with "once I see the semantic model in UML I know the ship is likely to sink."? | 18:05.12 |
| Robin_Watts | marcosw_: Where is the gs coverage stuff ? | 18:06.41 |
| marcosw_ | tkamppeter: I'm not really doing that; it was an attempt at humor. | 18:09.55 |
| | Robin_Watts: http://www.ghostscript.com/coverage/ghostpcl/ and http://www.ghostscript.com/coverage/mupdf/ (ghostscript is running now). | 18:10.34 |
| Robin_Watts | but where will gs be when it finishes? | 18:10.48 |
| | coverage/ghostscript/ ? | 18:10.58 |
| marcosw_ | yes, unless someone has a better suggestion. | 18:11.19 |
| Robin_Watts | marcosw_: I'm putting links on the dashboard now. | 18:12.12 |
| | http://ghostscript.com/regression/index2.html | 18:12.22 |
| ray_laptop | I came up with a "clever" (aka hackish) way to allow customer 801 select a TargetSize and have the input file be scaled (using Policy 13) to that size media. | 18:14.11 |
| | all using setpagedevice magic and some special Ghostscript extensions :-) | 18:14.32 |
| | now I have to go take a hot shower to feel clean again ;-) | 18:15.16 |
| henrys | tkamppeter: sorry I am not a big fan of UML, schemas or anything else that attempts to shortcut an english description. Those diagrams are taken straight from the XPS manual, which is an abomination. I think we've gone astray in our description of things, compare that with the classic Adobe Postscript Manual (red book) where things are described in well written prose. Usually the farther we get from that documentation paradig | 18:19.02 |
| | the higher the probability of failure (IMHO). The PDF manual has gotten much worse since adobe no longer is the owner. | 18:19.03 |
| | excuse my unsolicited ranting ... | 18:19.57 |
| tkamppeter | henrys, Adobe id not the owner of the manual any more? Or even not the owner of PDF any more? | 18:20.34 |
| Robin_Watts | Adobe wrote the PDF spec and maintained it up to 1.7. | 18:21.15 |
| | Then it became an international standard (ISO 93xxx or something) and the first thing they did was ruin the manual. | 18:21.49 |
| | The typesetting is worse now, certainly. The content will rot over time, I feel sure. | 18:22.17 |
| | The PDF reference manual describes some things poorly - such as patterns and their gstates for example. | 18:23.10 |
| | While I understand the drive to describe things formally (the desire to be able to reason about specifications logically is understandable, and laudable), history has shown that we are incapable of formulating such accurate descriptions. | 18:24.28 |
| vtorri_ | if pdf_load_page is called with a page number greater or equal the page count, does it return NULL ? | 18:25.06 |
| tkamppeter | Robin_Watts, is Adobe still developing on PDF or is PDF an independent (free?) standard now? | 18:26.06 |
| vtorri_ | +than | 18:26.11 |
| Robin_Watts | tkamppeter: It's an ISO standard. | 18:26.23 |
| | Adobe still seem to be driving it with extensions though. | 18:26.33 |
| | paulgardiner and I wrote a videophone in a previous life. The standard we were working to had a formally defined signalling system for call negotiation etc. | 18:27.15 |
| | So paulgardiner wrote some cunning code to generate the required C from the formal definition lifted directly from the standard. | 18:27.41 |
| | Then he spent ages fixing the broken formal definition so it worked. | 18:27.59 |
| mvrhel_laptop | thats wild | 18:28.45 |
| tkamppeter | Robin_Watts, so it would be better, when defining a standard, implementing and testing it at the same time. | 18:29.26 |
| Robin_Watts | tkamppeter: Right, but that's never the way things work. People always write some code, get it working, then write the specification from it. | 18:30.43 |
| | Knuth tried writing TeX in a manner where docs and code were all wrapped up together (tangle/weave?), I think. That didn't catch on, thank god. | 18:32.02 |
| marcosw_ | henrys: are the -dGraphicsAlphaBits=4 and -dTextAlphaBits=4 options supposed to work with GhostPCL? A large number of files generate errors. | 18:32.21 |
| henrys | marcosw_:incompatible with raster ops | 18:32.41 |
| Robin_Watts | My experience of Z is that it's awful too. | 18:32.59 |
| marcosw_ | you okay if I stop testing GhostPCL during the weekly -dGraphicsAlphaBits=4 regression test? | 18:33.32 |
| Robin_Watts | There is a role for formal systems, but you need an impressively large brain to make it work. | 18:34.04 |
| henrys | marcosw_:yes I'd like that | 18:34.08 |
| Robin_Watts | While I'm adding links to the dashboard, are there any more that people can think of? | 18:36.00 |
| henrys | Robin_Watts:a link to send an SMS to marcosw ;-) | 18:36.43 |
| Robin_Watts | henrys: That's doable, but we'd need to pay a subscription :) | 18:37.44 |
| marcosw_ | Robin_Watts: no you don't email to 510xxxxxxx@txt.att.net gets to me. | 18:39.21 |
| | I probably should remove that from the web copy of the irc logs :-) | 18:39.56 |
| Robin_Watts | marcosw_: I'm not sure that mentioning that in a logged public session was smart :) | 18:39.58 |
| | gawd. The curse of working with vmware. | 18:40.33 |
| | Start the VMware image... wait for the virtual machine to apply all the updates since you last used it... try to remember why you started it in the first place | 18:41.01 |
| henrys | then fail the "remember part" and shut if down is my work flow. | 18:48.20 |
| | s/if/it | 18:48.25 |
| marcosw_ | You guys are missing the important part of the meeting. We just discussed fax modems in the cloud. | 18:49.42 |
| tkamppeter | marcosw_, who is still using fax? For me it seems that fax will soon die. | 18:52.04 |
| | marcosw_, the biggest german telco, Deutsche Telekom (T-Mobile in the US), wants to remove all anoalog and ISDN network infrastructure until 2018 and have landline voice communication only via VoIP. AFAIR VoIP cannot transport fax. | 18:56.06 |
| Robin_Watts | tkamppeter: I think people would be very annoyed to find fax stopping working. There is no real alternative yet. | 19:01.11 |
| | a faxed document has legal standing in a way that an emailed document does not, IIRC. | 19:01.34 |
| marcosw_ | Robin_Watts: I remember that argument when it applied to Telex, that it would never replace fax since fax had no receive confirmation (apparently Telex does (or did, does Telex still exist?), if you received confirmation of a successful telex transmission that meant that not only had it been sent but that the the receiving machine printed it). | 19:12.59 |
| Robin_Watts | marcosw_: If I want to convert money from my dollar account into my euro account, I can do that by sending a faxed instruction to my bank. | 19:14.50 |
| | 1) because I have a decent bank :) and 2) because I have a fax mandate set up. | 19:15.04 |
| vtorri_ | if pdf_load_page is called with a page number greater or equal than the page count, does it return NULL ? | 19:15.18 |
| Robin_Watts | Without fax... how can I do that? Email is insecure. | 19:15.20 |
| | vtorri_: Either NULL or it may throw an error. | 19:15.36 |
| vtorri_ | Robin_Watts, the documentation does not say that it can throw an error | 19:16.05 |
| Robin_Watts | Well, then it probably returns NULL :) | 19:16.18 |
| vtorri_ | ok | 19:16.21 |
| marcosw_ | doesn't your bank have a web site? That's how I do most of my banking. | 19:16.31 |
| vtorri_ | i guess that all functions that return an error are correctly documented | 19:16.41 |
| | thanks | 19:16.45 |
| Robin_Watts | marcosw_: Yes it does, but for some things, you will always need to resort to something other than the web. | 19:16.56 |
| | (For instructions of suitable size/complexity) | 19:17.06 |
| | and I really don't want to go back to the days of having to write letters and post them. | 19:17.28 |
| marcosw_ | Robin_Watts: how does the bank know it's you sending the fax? | 19:18.23 |
| Robin_Watts | marcosw_: The incoming number. And my signature on the faxed document. | 19:18.46 |
| henrys | I thought signed and encrypted email was a pretty well understood/solved problem. | 19:22.05 |
| marcosw_ | is caller id secure in the UK? It's not in the US. and signatures on faxed documents are easy to forge, i.e. I'm pretty sure a scan of my signature would look authentic if it's faxed. | 19:22.43 |
| Robin_Watts | henrys: Email is pretty much as secure as a picture postcard. | 19:23.22 |
| marcosw_ | Robin_Watts: I just found that's it's illegal to spoof caller id in the UK, so it must be possible to do so. | 19:23.36 |
| Robin_Watts | There may be extensions etc to make you handle signing etc, but none of them are standard. | 19:24.15 |
| | and there are no solutions for securely exchanging keys automatically on email, right? | 19:24.36 |
| marcosw_ | Robin_Watts: by your argument as long as it had your return address and your signature on it a picture postcard would be secure. | 19:24.49 |
| Robin_Watts | so until it's to the stage where 90% of the population are using it without realising it, it's not a solution. | 19:25.23 |
| henrys | Robin_Watts:yes a third party is needed ⦠by your argument you shouldn't be using amazon. | 19:25.26 |
| vtorri_ | what does fz_bound_page return ? | 19:25.43 |
| Robin_Watts | henrys: no, my argument can be boiled down to this... could my mother in law use it? | 19:25.54 |
| vtorri_ | it returns a fz_rect * | 19:25.55 |
| Robin_Watts | My mother in law can use a fax. | 19:25.59 |
| vtorri_ | but what is it ? | 19:26.07 |
| Robin_Watts | My mother in law can use a phone. She can just about use email. She can barely use the web - certainly she wouldn't do internet banking. | 19:26.48 |
| | Until there is an alternative in place for her, fax can't be allowed to die. It's why we still have cheque books. | 19:27.22 |
| henrys | oh I thought you were saying there was a technological barrier to having secure email. | 19:27.26 |
| Robin_Watts | henrys: No, there is no technological barrier. There is an implementation barrier - no one has done it in an accessible way yet. | 19:28.40 |
| | marcosw_: There are 2 issues; 1) is it really from me? and 2) can anyone read it as it's delivered. | 19:29.15 |
| | 1) is solved by the bank calling me back to verify it's me (they've done that in the past) | 19:29.37 |
| | 2) is not possible with a picture postcard. | 19:29.55 |
| | henrys: Do we have an nda in place with zeniko? | 19:35.05 |
| henrys | I think we only have the CLA in place. | 19:35.30 |
| marcosw_ | Robin_Watts: if the picture postcard has your return address and signature on it it must be for you. At least that's what you said about a fax. It's true that a return address is easier to spoof than caller id, but as I've pointed out a faxed signature is easier to forge than a signature on a postcard. | 19:35.32 |
| Robin_Watts | If we want to enlist his help in the OpenJPEG fuzzing fixes, we'll need to share the fuzzing files with him. Do we need an NDA for that? | 19:36.03 |
| henrys | that's the next thing to get rid of after faxes -- the postal service. | 19:36.08 |
| Robin_Watts | marcosw_: With a picture postcard, you can't phone the sender up to verify that they sent it. With a fax you can. | 19:36.41 |
| henrys | Robin_Watts: what I read of your discussion with the client , yes, did you have a different interpretation? | 19:36.48 |
| Robin_Watts | henrys: I think it's fine to share the stuff with him, as long as he promises not to forward them on. | 19:37.24 |
| henrys | okay that suits me as well. | 19:37.41 |
| marcosw_ | Robin_Watts: huh? If I send you a postcard you can't call me and ask me if I sent it to you? | 19:38.07 |
| Robin_Watts | marcosw_: Not easily, because you aren't likely to be stood by the phone after however long it's taken the postcard to arrive ;) | 19:38.50 |
| | marcosw_: The point is, that the banks here consider that the combined difficulty of 1) forging the fax return number, 2) forging the signature, 3) passing any other security tests they might have in place is sufficient for them to accept fax instructions. | 19:40.16 |
| marcosw_ | Robin_Watts: is jpeg/png reading in mupdf working yet? mvrhel_laptop points out it's on the slides | 19:41.16 |
| Robin_Watts | marcosw_: It is in and working, yes. | 19:41.27 |
| | jpeg/png/tif. | 19:41.31 |
| marcosw_ | thx | 19:41.45 |
| mvrhel_laptop | ok so you can bascially do image conversions now . can you go to pdf from these? | 19:42.04 |
| | i.e. is there a pdfwrite device working? | 19:42.18 |
| | Robin_Watts: ^^ | 19:42.22 |
| Robin_Watts | I have a prototype pdfwrite device. | 19:43.04 |
| mvrhel_laptop | ok | 19:43.08 |
| Robin_Watts | It is sufficient for tiger. | 19:43.10 |
| | I can't remember if it does images. | 19:43.20 |
| | it doesn't do fonts, certainly. | 19:43.30 |
| | svgwrite is more advanced; most things except fonts are working. | 19:43.50 |
| mvrhel_laptop | ok thanks | 19:44.21 |
| henrys | mvrhel_laptop: you are slotted for 1:30 to 5:00 ⦠is Richard Hughes of red hat going first. | 19:49.53 |
| | ? | 19:50.04 |
| mvrhel_laptop | henrys: I think it had me first, then richard, then til | 19:50.53 |
| henrys | why don't they have audio in the webex thing? | 19:59.51 |
| marcosw_ | henrys: be glad they have the slides on webex; Robin_Watts suggested they fax the slides to people who are offsite :-) | 20:02.59 |
| Robin_Watts | Hmm. My phone has a speakerphone mode, but no secrecy mode. So I can't dial in for the talk as otherwise everyone would hear the dogs barking :( | 20:03.57 |
| tkamppeter | henrys, mvrhel_laptop, marcosw: I remember that Richard was first but I am not sure. I will do mine last. | 20:09.21 |
| | Robin_Watts, no microphone mute on your phone? No wife or kids at home who keep the dogs distant from you? | 20:10.23 |
| | henrys, are you calling in? | 20:13.12 |
| henrys | no I was going to watch the slides go by and standby on irc - marcosw will alert us if something comes up | 20:14.11 |
| vtorri_ | tor8, what does pdf_bound_page() return ? | 20:15.04 |
| tor8 | vtorri_: the page size | 20:15.35 |
| vtorri_ | which is also in the 2nd parameter, right ? | 20:15.53 |
| | third | 20:16.13 |
| | not 2nd | 20:16.17 |
| henrys | marcosw_:last I looked on webex it seemed like michael sweet's stuff was going to go over into the next time slot | 20:16.23 |
| tor8 | you probably should be calling fz_bound_page unless you have a very good reason | 20:16.41 |
| | yes, it returns the third argument after filling it out. for convenience. | 20:17.05 |
| vtorri_ | tor8, well, actually, i don't know the differences between the fz_ functions and the pdf_ functions | 20:17.15 |
| Robin_Watts | tkamppeter: No microphone mute on the downstairs phone. | 20:17.17 |
| tor8 | so you can call fz_transform(..., fz_bound_page(..., &bbox)) | 20:17.20 |
| Robin_Watts | and the one in my office is inconvenient for this time of night. | 20:17.36 |
| tor8 | vtorri_: you shouldn't have to use the pdf_ functions at all. everything you need should be covered by the functions that relate to fz_document | 20:17.58 |
| henrys | Robin_Watts: thanks for sticking around I know it is late for you. | 20:18.34 |
| vtorri_ | so what are those pdf_ functions for ? (i'm just curious) | 20:18.35 |
| marcosw_ | henrys: they started late this morning since they had some stuff they didn't get to last night. the 1:00 session will start at 130 | 20:19.02 |
| tor8 | vtorri_: they are called by the fz_ functions if the document is a pdf. there are also xps_ and cbz_ versions of the same functions. | 20:19.04 |
| vtorri_ | ok | 20:19.28 |
| tor8 | the fz_ functions take care of hiding the details of which exact format is used, so you only need to write your code once for all formats | 20:19.38 |
| henrys | marcosw_:who knows when the clouds will clear | 20:19.43 |
| | ? | 20:19.44 |
| vtorri_ | tor8, so to manage pdf, xps and cbz docments, i have to use the fz_ functions, right ? | 20:19.53 |
| | ok | 20:20.02 |
| | thanks | 20:20.06 |
| tor8 | vtorri_: yes. | 20:20.14 |
| vtorri_ | good to know :) | 20:20.16 |
| hughsie-afk-mtng | tkamppeter, hey :) | 20:20.59 |
| vtorri_ | but i have to fix my lib, then :) | 20:21.18 |
| tkamppeter | henrys, according to the OpenPrinting weather forecast it should be at 1:30pm, as we lost 30 min in the morning for the GNOME/GTK Q&A. | 20:21.20 |
| | hughsie-afk-mtng, what does the mtng in your nick mean? | 20:21.52 |
| hughsie-afk-mtng | tkamppeter, meeting | 20:22.04 |
| | if i drop the afk i get people asking me questions.... | 20:22.34 |
| vtorri_ | tor8, what's the difference between fz_pre_scale() and fz_scale() ? the doc is not giving much informations | 20:22.46 |
| tkamppeter | Always nice when they are afk and typing, how far is it with speech input on PCs? | 20:22.48 |
| tor8 | fz_scale: Create a scaling matrix. | 20:23.22 |
| | fz_pre_scale: Scale a matrix by premultiplication. | 20:23.29 |
| | what is unclear? | 20:23.34 |
| vtorri_ | tor8, "premultiplication" :) | 20:23.47 |
| tkamppeter | hughsie-afk-mtng, yes, then #ghostscript will get a CM support channel ... | 20:23.50 |
| | ... but mvrhel_laptop could perhaps help answering then. | 20:24.10 |
| tor8 | fz_scale: M = [sx 0 0 sy] | 20:24.20 |
| | fz_pre_scale: M = [sx 0 0 sy] * M | 20:24.33 |
| vtorri_ | ok, thanks | 20:24.51 |
| hughsie-afk-mtng | tkamppeter, i'm still trying to work through all the super lcms transform stuff the gs guys shared with me last week... :) | 20:24.59 |
| tor8 | (or the other way round, never can get matrix multiplication order correct) | 20:25.06 |
| vtorri_ | :) | 20:25.16 |
| tkamppeter | hughsie-afk-mtng, what super lcms stuff. | 20:25.41 |
| vtorri_ | tor8, maybe this should be added in the doc | 20:25.54 |
| tor8 | it could probably be clarified more, yes | 20:26.21 |
| hughsie-afk-mtng | tkamppeter, see my posts to the lcms mailing list -- basically, i wanted to use threads in colord to accelerate a color transform, but the gs guys pointed out that the transform should probably be optimised as there are some big gains | 20:26.40 |
| tkamppeter | hughsie-afk-mtng, mvrhel_laptop, was there no work on the GS side to make lcms thread-safe. | 20:27.49 |
| hughsie-afk-mtng | tkamppeter, my understanding is that lcms is thread safe if you use it correctly | 20:28.22 |
| mvrhel_laptop | you just cant share transforms nor profiles amongst threads | 20:28.49 |
| tkamppeter | it seems that the clouds are clearing in a few minutes. | 20:28.59 |
| | hughsie-afk-mtng, seems that its your turn now, good luck! | 20:29.21 |
| hughsie-afk-mtng | tkamppeter, can you load up my slides on the projector pls | 20:29.38 |
| robin_watts_mac | lcms 2 is thread safe (in the latest version) | 20:29.51 |
| | lcms 1 is not. | 20:29.56 |
| | older versions of lcms 2 have problems. | 20:30.07 |
| hughsie-afk-mtng | sorry, i meant lcms2 | 20:30.12 |
| tkamppeter | hughsie-afk-mtng, mike is doing this stuff, my arms are not long enough. | 20:30.15 |
| hughsie-afk-mtng | kk | 20:30.22 |
| sebras | hi! let me ask an xps-newbie question: is it possible to easily convert xps to pdf losslessly? | 20:33.54 |
| robin_watts_mac | sebras: PDF can probably express everything that XPS can. | 20:34.47 |
| | but the only conversion route I know of currently is gs and pdfwrite | 20:35.09 |
| sebras | robin_watts_mac: ok.this is mostly paths and text, so it should be fine. | 20:35.11 |
| tor8 | define losslessly | 20:35.13 |
| robin_watts_mac | and I wouldn't strictly call pdfwrite lossless. | 20:35.26 |
| sebras | tor8: too hard. :-) | 20:35.30 |
| mvrhel_laptop | only issue is deviceN colors from xps which use ICC profiles where pdf is limited to gray, rgb and cmyk profiles | 20:35.33 |
| robin_watts_mac | but for paths and text you should be fine. | 20:35.35 |
| sebras | mvrhel_laptop: color accuracy is not important. | 20:35.48 |
| tor8 | there are things in xps that can't be perfectly represented in pdf | 20:35.50 |
| | like unicode text | 20:36.02 |
| sebras | tor8: oh. | 20:36.11 |
| robin_watts_mac | different mitring rules. | 20:36.11 |
| | and different dashing rules. | 20:36.27 |
| tor8 | xps has a separate unicode and glyph index arrays in their text objects | 20:36.40 |
| | pdf has only one "array" that gets converted by separate encoding mechanisms to get glyphs and unicode text out | 20:37.04 |
| | xps has triangle line caps, which is missing from pdf | 20:37.34 |
| sebras | tor8: alright, but if I use gs to convert a text-only xps to pdf then I won't mess up the encoding, normally, right? | 20:37.50 |
| tor8 | xps has transparency in shadings, which is missing from pdf (but can be emulated with soft masks) | 20:37.53 |
| sebras | tor8: enough! :-) | 20:38.04 |
| tor8 | xps with text -> pdf with text will get you reasonable results | 20:38.11 |
| | images may or may not transfer losslessly, kens would know more about that | 20:38.39 |
| sebras | tor8: excellent. and -sDEVICE=pdfwrite appears to be my friend. I'll try it. | 20:38.41 |
| | tor8: images are not terribly important, so that's ok. | 20:38.54 |
| tor8 | there are a few scenarious where (in the past, most are fixed now I think) pdfwrite would fall back to dumping rasterized bitmaps of the page into the pdf | 20:39.46 |
| sebras | pdfwrite with xps-input work well. thanks! | 21:03.47 |
| robin_watts_mac | has michael done his talk yet ? | 21:20.19 |
| marcosw_ | not yet, he's about to start. | 21:20.30 |
| robin_watts_mac | ok. I have the lappy by the telly, and am following slides. | 21:20.47 |
| mvrhel_laptop | slides are going to fly by fast now.... | 21:21.21 |
| marcosw_ | mvrhel_laptop: is starting his talk | 21:22.15 |
| | robin_watts_mac: we are doing the mupdf portion of the presenation | 22:18.19 |
| henrys | wow he's already 1 hour into this. does he have water? | 22:19.21 |
| robin_watts_mac | yeah, watching it along on webex. | 22:25.36 |
| henrys | marcosw_:lots of questions? | 22:31.59 |
| marcosw_ | nothing interesting. Questions about mupdf vs ghostscript, multi-threaded rendering, size of mupdf. | 22:41.19 |
| | til and mike sweet are asking about add cups output to mupdf | 22:41.53 |
| | also til asked about a performance comparison of popplar to mudf. | 22:42.16 |
| robin_watts_mac | cups output would be easy to add. | 22:42.33 |
| | it's just a different bitmap format. | 22:42.47 |
| | we'd love to see a performance comparison to poppler :) | 22:43.10 |
| marcosw_ | someone is asking saying that there is a vertical writing bug in mupdf, we are asking him to enter a bug in the tracker | 22:45.18 |
| robin_watts_mac | marcosw: I fixed one last week. | 22:45.39 |
| | so depending on when he tried it, it might be fixed already. | 22:46.28 |
| vtorri_ | i've once written my lib with poppler and mupdf as backend | 22:48.50 |
| | mupdf was way faster on smalll devices | 22:49.05 |
| marcosw_ | presentation done. we are one break befor tkamppeter speaks. | 22:49.20 |
| robin_watts_mac | vtorri_: How long ago was this? MuPDF should have gotten faster recently. | 22:49.35 |
| henrys | applause! great job mvrhel_laptop | 22:49.46 |
| vtorri_ | robin_watts_mac, i've dropped poppler support :) | 22:49.53 |
| | 2 years ago | 22:50.00 |
| robin_watts_mac | mvrhel_laptop: yeah, nice one! | 22:50.01 |
| vtorri_ | something like that | 22:50.07 |
| | maybe more | 22:50.10 |
| robin_watts_mac | vtorri: ah, so we should be even better now. thanks! | 22:50.15 |
| vtorri_ | i'm not interested in poppler anymore anyway | 22:50.22 |
| | poppler break API on layers that should keep stable API | 22:51.43 |
| | and it's c++ | 22:51.48 |
| mvrhel_laptop | thanks. it all went pretty well. | 22:52.53 |
| henrys | marcosw_, mvrhel_laptop I have an appointment and will have to miss till's presentation. | 22:53.06 |
| robin_watts_mac | and I have an appointment with my bed :) | 22:55.49 |
| | I'll review the slides tomorrow. | 22:56.10 |
| mvrhel_laptop | robin_watts_mac: you may want to look at til's slides from yesterday | 22:56.58 |
| | apparently to get mupdf into the running for ubuntu touch we would need to get some filters implemented | 22:57.20 |
| robin_watts_mac | I'll look through all of them. | 22:57.21 |
| mvrhel_laptop | tkamppeter: you may want to chat with robin_watts_mac about this tomorrow | 22:57.39 |
| robin_watts_mac | ah, i think we've spoken about that. | 22:57.46 |
| mvrhel_laptop | oh good. | 22:57.53 |
| Robin_Watts | I have mupdf running on windows 8 with the new locking. | 23:02.47 |
| mvrhel_laptop | great! | 23:02.56 |
| | I got the progress bar working for text search. but something odd is going on in my simulator | 23:03.23 |
| Robin_Watts | It was marvellously straightforward. | 23:03.26 |
| tkamppeter | mvrhel_laptop, what is happening in the room, we are at the hour now. | 23:03.41 |
| mvrhel_laptop | the flipview is locking up now and then | 23:03.47 |
| Robin_Watts | The hardest thing was having to install git and clone it locally cos it won't run from a network share. | 23:03.59 |
| mvrhel_laptop | tkamppeter: people standing around chatting | 23:04.02 |
| tkamppeter | mvrhel_laptop, flipview? | 23:04.06 |
| mvrhel_laptop | tkamppeter: that was for Robin_Watts | 23:04.13 |
| tkamppeter | mvrhel_laptop, sorry. | 23:04.26 |
| mvrhel_laptop | tkamppeter: no worry my fault for not adding his name | 23:04.36 |
| tkamppeter | mvrhel_laptop, so tell them to continue the session. | 23:04.38 |
| mvrhel_laptop | let me get mike and ira | 23:04.50 |
| | here they come | 23:04.55 |
| Robin_Watts | mvrhel_laptop: How can I zoom ? | 23:09.24 |
| mvrhel_laptop | Robin_Watts: are you in the simulator | 23:10.04 |
| | if not, I just realized I need to add keyboard option.... | 23:10.23 |
| Robin_Watts | I'm on Windows 8. | 23:10.30 |
| mvrhel_laptop | Robin_Watts: I need to add keyboard zooming. I only have touch zooming | 23:10.57 |
| Robin_Watts | right. | 23:11.04 |
| | same as android then. | 23:11.08 |
| mvrhel_laptop | I am currently seeing some update issues in the simulator. need to see if they occur on the actual device | 23:12.18 |
| | tomorrow | 23:12.21 |
| Robin_Watts | yeah, tomorrow for me too. Night! | 23:12.42 |
| mvrhel_laptop | good night. thanks for staying up | 23:13.24 |
| Robin_Watts | no worries. | 23:13.33 |
| | Forward 1 day (to 2013/05/16)>>> | |
IRC Logs