| <<<Back 1 day (to 2012/09/30) | 2012/10/01 |
fgro | how can i convert a pdf (that is the first page of the pdf) to an image from commandline? | 08:31.03 |
kens | What kind of image ? | 08:31.17 |
fgro | png/jpg would be fine | 08:31.28 |
kens | gs -sDEVICE=jpegrgb -o out.jpg -dLastPage=1 input.pdf | 08:31.54 |
fgro | kens: cheers! | 08:32.07 |
Robin_Watts | Use png though. | 08:32.17 |
| png16m. | 08:32.19 |
fgro | very nice | 08:32.39 |
Robin_Watts | gs -sDEVICE=png16m -o out.png -dLastPage=1 input.pdf | 08:32.43 |
| If you have any (readable) text on the front page, then png is a much better bet than jpeg. | 08:33.24 |
fgro | i was looking into imagemagicks convert command at first, but it creates an image for every page and there doesnt seem to be a way to limit. than i read that "convert" uses "gs" internally ... much better. thx! | 08:33.54 |
| oh. is there a way to "autodetect" first page with "readable" content? | 08:34.25 |
Robin_Watts | frgo: No. | 08:34.36 |
kens | No, there's no way to know what is readable | 08:34.40 |
fgro | ok thanks! | 08:34.46 |
| one last question, if i want to convert multiple pages, say the first 5, each into a single file? | 08:38.06 |
fgro | is reading gs man page... still its a bit confusing... | 08:38.33 |
kens | fgro use -dFirstPage= and -dLastPage= to specify a contiguous range of pages | 08:40.38 |
| use -o out%d.png to get a separate file for each page | 08:40.56 |
fgro | kens: fantastic! | 08:41.23 |
kens | We don't write the man pages, you'd bbe better off with the Ghostscript documentation which is written in HTML and in gs/doc | 08:41.32 |
fgro | kens: ok. thanks again! | 08:43.36 |
kens | chrisl ping | 10:38.54 |
chrisl | kens: pong | 10:43.02 |
Robin_Watts | sebras: OK, looking at your patches now. | 11:02.30 |
kens | chrisl how do you build GS on Linux without X11 ? | 11:02.52 |
Robin_Watts | --with-no-x11 ? | 11:03.35 |
| --with-no-x=yes I think. | 11:04.03 |
kens | I don't know, was hoping chris would | 11:04.16 |
chrisl | I think it's configure --without-x | 11:04.33 |
kens | OK that sounds right, thanks | 11:04.42 |
chrisl | Yes, that seems to work | 11:06.13 |
kens | Thanks chrisl | 11:06.30 |
chrisl | Robin_Watts: I didn't post that empty message on Saturday - strange occurance. | 11:06.50 |
Robin_Watts | I blame your cat. | 11:07.08 |
chrisl | She was sleeping next to the keyboard when I got home, but probably still a stretch.... | 11:07.42 |
| <sigh> dying hard disk + Windows XP == less than happy me :-( | 11:08.30 |
kens | Parental problem ? | 11:09.10 |
chrisl | Another relative - my aunt's PC. | 11:09.25 |
kens | close enough.... | 11:09.32 |
Robin_Watts | chrisl: http://www.amazon.co.uk/dp/B001VSQR98/?tag=hydra0b-21&hvadid=10339759509&ref=asc_df_B001VSQR98 | 11:09.56 |
chrisl | I've been tempted! | 11:10.28 |
Robin_Watts | I had that before, but I think Helen binned it. | 11:10.48 |
chrisl | *very* strange: bios reports a hard disk error, takes several tries to boot Windows, once Windows is running, everything is (apparently) fine. | 11:11.00 |
Robin_Watts | ponders a womens version that says "No, I cannot fix my computer." | 11:11.42 |
kens | Oooh, sexist.... | 11:11.55 |
Robin_Watts | :) | 11:12.00 |
chrisl | With some, I feel more appropraite would be: "No, I *will* not fix my computer." | 11:12.25 |
| Some people seem wilfully stupid when confronted with a keyboard and monitor...... | 11:12.55 |
Robin_Watts | "Why should I fix my computer, when you can do it for me?" | 11:12.58 |
chrisl | Yes, we have set a poor precedent :-( | 11:14.21 |
Robin_Watts | tor8: ping | 11:24.37 |
tor8 | hi robin | 11:24.48 |
Robin_Watts | sebras points out in one of his commits that we use the terms axial and linear interchangably w.r.t shadings. | 11:25.01 |
| He proposes moving to use axial everywhere. | 11:25.08 |
| I agree, except I prefer linear. | 11:25.17 |
| Do you have a feeling on this? | 11:25.29 |
tor8 | oh. right. probably my confusion between xps and pdf terminology. | 11:25.40 |
| I think I also prefer linear | 11:25.47 |
Robin_Watts | OK. Ta. | 11:26.00 |
| tor8: I've just updated my master branch with all sebras patches on top of my shadings stuff. | 11:49.00 |
| I've deliberately done all the changes on a branch which I've then merged back down, as it means the customer who is interested in this, only needs to pull in a single commit. | 11:49.37 |
| Are you happy with sebras having reviewed it, or do you want to look yourself? | 11:49.56 |
| (re cluster testing now) | 11:50.07 |
| lunchtime. | 11:51.18 |
sebras | reads logs. | 12:45.15 |
| tor8: I have no opinion on linear/axial, just that we pick one term and stick with it. :) | 12:46.05 |
| Robin_Watts: I think that tor8 ought to have look too. I've only skimmed certain parts that are hard for me to understand -- notably the edgelist stuff. I'm not even sure that this has changed. | 12:47.05 |
Robin_Watts | It hasn't. | 12:47.17 |
sebras | Robin_Watts: right, then that's not an issue at least. :) | 12:48.28 |
| what about moving the trig functions? | 12:48.41 |
Robin_Watts | I think I'll do that, but not on the shading branch. | 12:48.56 |
sebras | why not rebase the shading branch on top of master? | 12:49.28 |
| the customer will still only need to keep track of a single sha1... | 12:49.44 |
Robin_Watts | because I have a customer who is interested in this, but really wants to just cherry-pick a single commit. | 12:49.50 |
| This way he can cherry pick the merge commit and get all the shading commits in one go, I think. | 12:50.15 |
sebras | ah, so they want it based on an older version then I guess. | 12:50.17 |
Robin_Watts | And I like keeping the history. | 12:50.20 |
| yeah. | 12:50.22 |
sebras | I'm not saying rebase and squash on top of master, just rebase the entire patch series. | 12:50.40 |
| but if the customer wants to cherry pick just a single commit then I can see why a merge commit makes more sense. :) | 12:51.06 |
Robin_Watts | sebras: That would ordinarily be what I'd do, but then we lose the "1 cherry-pick"-ness. | 12:51.10 |
sebras | mmm. | 12:51.16 |
| this is what I do all the time. I think this makes the version tree look saner. | 12:51.32 |
Robin_Watts | It's also what I do all the time. | 12:51.42 |
sebras | also it prevents me from forgetting old branches. :) | 12:51.49 |
| alright, good work. now we'll just have to deal with the FIXMEs sometime. ;) | 12:52.28 |
Robin_Watts | OK. We get 680 diffs with the shadings, all look just like small rounding changes. | 12:52.40 |
sebras | mmm, this is what I saw when running sane too. | 12:52.56 |
| I have to admit thought that I can't spot where the rounding changes were introduced though. | 12:53.29 |
Robin_Watts | We used to transform at load time. | 12:54.33 |
| We now transform at render time. | 12:54.38 |
sebras | Robin_Watts: are you thinking about ctm? | 13:07.40 |
| Robin_Watts: hm.. yes, I guess that the order of the multiplications for the matrices might actually be the cause. | 13:08.12 |
Robin_Watts | sebras: yeah | 13:13.24 |
| the more I think about it, the more I am tempted to just rebase it onto the branch. | 13:16.56 |
| If the customer wants a single commit, I can always fake that for them later. | 13:17.18 |
| s/the branch/master/ | 13:17.29 |
sebras | mmm, if this is just a one off for the customer you could just squash the commits and provide the customer with a diff-file? | 13:25.12 |
| I'm not sure how you guys normally handle things like this. :) | 13:25.30 |
| anyway, the code itself is good, and assume you were happy with my fixes too. | 13:25.59 |
Robin_Watts | sebras: Yeah, I could make a customerXXX-shading tag with the relavent commits squashed on. | 13:26.02 |
| sebras: absolutely. | 13:26.07 |
sebras | cool. | 13:26.11 |
Robin_Watts | tor8, sebras: I fancy the idea of exposing pdf_function as fz_function. | 13:26.59 |
sebras | Robin_Watts: why? | 13:27.13 |
| are they present in xps? | 13:27.18 |
Robin_Watts | The way I'd do it is to add an 'evaluate' function pointer to the structure, and then do each different pdf type as an instance of it. | 13:27.52 |
| sebras: It would mean that shadings could pass the function without having to sample it. | 13:28.19 |
tor8 | Robin_Watts: ah, but we sample it for performance don't we? | 13:28.43 |
| well, partly at least | 13:28.51 |
Robin_Watts | tor8: do we? | 13:28.54 |
| At the moment we sample it at a 32x32 frequency, which for many renderings is way more than we need. | 13:29.13 |
tor8 | we render in t-space and then use the sample lookup table to convert to color | 13:29.14 |
Robin_Watts | I'm thinking of the 'function based' ones. | 13:29.39 |
| type1 shadings. | 13:29.51 |
tor8 | Robin_Watts: oh, those | 13:30.01 |
| that's a different use of functions yet again | 13:30.20 |
Robin_Watts | At the moment they take 32*32*4bytes*number of components bytes in each shading. | 13:30.32 |
tor8 | wouldn't it be possible to represent type 1 shadings as one (1) quad in the mesh? | 13:32.20 |
| wonder why we sample it at 32*32 | 13:32.30 |
| is it because they can have two levels of function lookups? | 13:32.42 |
Robin_Watts | functions can be discontinuous; meshes can't? | 13:33.26 |
| Or single quads within a mesh can't. | 13:33.35 |
| In the new code, the function based shadings end up as quads in the same way that the rest of the shadings do. | 13:34.17 |
| sebras: You were asking about the order of winding of points in a quad yesterday. | 13:35.38 |
| Draw yourself a square and label the corners in clockwise order, 1,2,3,4 | 13:36.02 |
| Then draw a line from 2 to 4. | 13:36.08 |
| We break that quad down into 1,2,4 and 2,3,4 | 13:36.31 |
| Sorry. | 13:37.09 |
| 2,4,3 not 2,3,4 | 13:37.19 |
| I believe that this is important because that way both triangles have one edge that runs from 2 to 4, hence both should hit exactly the same points. | 13:38.17 |
| Likewise, if you consider the edges shared with neighbouring triangles (from adjacent quads in the mesh) we always walk those edges in the same direction too. | 13:38.51 |
sebras | Robin_Watts: right, but as I understood it the direction of the triangles in the middle of the quad should be in opposite directions, no..? | 13:43.56 |
| I'm not sure I remember _why_ that is the case though. :) | 13:44.13 |
Robin_Watts | sebras: Well, they kind of are. | 13:44.55 |
| Any point within the 1,2,3 triangle is encircled in a clockwise triangle. Any point within the 2,4,3 is encircled anticlockwise. | 13:45.32 |
tor8 | Robin_Watts: zebras: the winding shouldn't matter to the triangle rasterizer | 13:46.24 |
sebras | d'oh, why didn't I just draw this before responding like I always do? I _never_ manage to do this correctly without it! :) | 13:46.26 |
| ztor8: is the winding only interesting in the case where the tri would be transparent otherwise? | 13:47.35 |
| or rather -- missing completely. | 13:47.41 |
| anyway I think I fixed the assumptions that were changes in linear/radial shadings yesterday so I believe it to be correct now. | 13:48.09 |
Robin_Watts | I believe in transparent cases, we always plot with alpha = 1 to a temporary mask and then blend, to avoid edge issues. | 13:48.30 |
sebras | and 2.7.1 and 2.7.3 pdfs are rendered correctly so I'm happy. | 13:48.46 |
| I shouldn't have said transparent because that has a different meaning in this context. argh! I need to learn the terminology. :-( | 13:49.18 |
| tor8: save me! | 13:49.31 |
| Robin_Watts: when rasterizing triangles in 3D graphics winding order matters as you probably know. if the order is right you see the triangle if it's wrong you don't. back-face culling is a term that comes to mind, but I'm not convinced it's the right one. | 13:52.51 |
| but if winding order doesn't matter to the triangle rasterizer then theres no problem. :) | 13:53.10 |
Robin_Watts | sebras: Ah, backface culling. | 13:53.53 |
| Yeah, we don't do that. | 13:53.57 |
| backface culling seems like the right term to me. | 13:54.18 |
sebras | good. :) | 14:08.35 |
| why we sample at 32x32 I don't know. I believe that if you git blame you'll find that this code has been in mupdf since shadings were introduced. | 14:09.00 |
| I don't remeber changing it at least. :) | 14:09.09 |
Robin_Watts | It's not an unreasonable setting; if we want to allow people to zoom in, we probably need to offer at least that much accuracy. | 14:09.38 |
| But it would be much nicer to not have to pick such an arbitrary size at load time. | 14:10.02 |
| now we have stuff done at the render time, we can be smarter about subdivision. | 14:10.28 |
sebras | I do see your point. | 14:10.39 |
Robin_Watts | I am pulling in a commit at the moment from my pdfwrite work. | 14:10.47 |
| In order to be able to write out images without recompression, I'd update the stream handling code so that we could open a stream from a pdf file with an fz_image_params *. | 14:11.32 |
tor8 | Robin_Watts: cps shadings have an alpha channel | 14:11.56 |
| s/cps/xps/ | 14:12.00 |
| damn autocorrect! | 14:12.12 |
Robin_Watts | If such a pointer is given, then we wouldn't actually instantiate the filter for the last step of decompression, we'd just write the details for that filter into the structure. | 14:12.21 |
| That meant that I could read the image data into a buffer still compressed. | 14:12.35 |
tor8 | stupid mountain lion keeps forgetting I tell it not to correct my spelling... | 14:12.36 |
Robin_Watts | I want to use that same code for mesh streams. | 14:13.02 |
| The only problem with doing that is the name of the structure. fz_image_params. | 14:13.28 |
sebras | Robin_Watts: fz_stream_params? :) | 14:21.45 |
Robin_Watts | sebras: Yeah, just toying with a refactoring now. | 14:22.03 |
tor8 | Robin_Watts: there's an odd ":wq" incantation at the end of one of the robin/master commits :) | 14:54.55 |
Robin_Watts | That's a sebrasism. | 14:55.49 |
tor8 | also, fz_shade.u.a_or_r shouldn't that be changed to l_or_r in the "linear" rather than "axial" commit | 14:56.18 |
Robin_Watts | tor8: good point. will do. | 14:56.31 |
ray_laptop | BTW, Adobe refers to Axial shadings. IMHO, linear is confusing since that can be confused with linear functions | 15:05.06 |
Robin_Watts | ray_laptop: Adobe does. XPS refers to linear gradients. | 15:06.04 |
| so either way we're going to be wrong for one of them. But both tor8 and I prefer 'linear', so... | 15:06.34 |
kens | rayjj : or anyone else. Is it actually *possible* in the current code to pass a dictionary as an argument to putdeviceparams ? There is a dictionary type defined, but if I try to use one, I get a seg fault. | 15:19.01 |
Robin_Watts | kens: I have a vague memory of people using dictionaries for setting up separation names etc for tiffsep etc? | 15:19.48 |
kens | Separation names would be a page device parameter wouldn't it ? | 15:20.14 |
| Therefore handled in PostScript. | 15:20.21 |
| I'm talking about putdeviceparams | 15:20.30 |
| Or more accurately .putdeviceparams | 15:21.19 |
Robin_Watts | I was thinking of SeparationColorNames, and SeparationOrder, which are (apparently) arrays anyway. | 15:22.12 |
kens | I must admit I'm puzzled as to whether this works | 15:22.29 |
| I have a nasty suspicion it doesn't | 15:22.43 |
Robin_Watts | tor8: New version of patches pushed. | 15:24.15 |
henrys | kens:I have looked for that in the past and I'm almost certain we don't have it. | 15:24.22 |
kens | henrys I was beginning to fear that was the case | 15:24.35 |
Robin_Watts | (just the a_or_r to l_or_r change, plus the compressed data stream refactor) | 15:24.44 |
kens | THere is code that looks like it ought to work, but I don't think it does :-( | 15:24.54 |
henrys | marcosw:so autogen.sh doesn't work for you? | 15:25.04 |
marcosw | henrys: right. | 15:25.30 |
| when I upgraded to mountain lion I had to install autoconf separately. | 15:26.02 |
chrisl | Hmm, I've always had to install autoconf separately on OS X, i think..... | 15:26.46 |
kens | I had to install autotools | 15:27.05 |
henrys | marcosw:I installed autoconf it is in /opt/local/bin/autoconf and it works for me. | 15:27.16 |
| maybe a path issues? | 15:27.38 |
marcosw | probably. Why not /usr/local/bin? | 15:28.55 |
henrys | because that is where macports installs it? | 15:29.14 |
| are you getting autogen another way? | 15:29.54 |
marcosw | I believe I installed it from source. I don't use macports | 15:30.12 |
henrys | I guess macports moved to /opt - it did work before and I usually use macports though I might have done something else with autoconf years ago. | 15:31.46 |
chrisl | It may be that the macports install only sets up paths for the current user (at least by default) | 15:32.47 |
henrys | no I have to set up the /opt/local/bin on my path, I've always had it, it's used by other stuff. | 15:35.02 |
| first paragraph here:http://guide.macports.org/chunked/installing.macports.html | 15:35.28 |
marcosw | henrys: I've added /opt/local/bin to my path, so everything appears to be working now. | 15:35.54 |
henrys | kens:why did you have to install git? xcode has support now? | 15:37.24 |
Robin_Watts | xcode has command line git in it ? | 15:37.56 |
kens | henrys I've never used XCode before, I had no idea it could work with a Git repository | 15:38.04 |
tor8 | Robin_Watts: fz_compressed_buffer sounds better than yoda wording does it not? | 15:39.04 |
Robin_Watts | tor8: if so you think. | 15:39.20 |
| change I will. | 15:39.27 |
henrys | well it is part of xcode and lives in /usr/bin/git so you shouldn't have had to install it. Not important just curious if you found something broken with the mac's git, if so I'll install the macports version. | 15:39.50 |
tor8 | Robin_Watts: kens: git comes with the xcode command line tools, and xcode 4 has some git integration (but I've never trusted Xcode 4 enough to actually rely on it) | 15:40.23 |
Robin_Watts | henrys: I have no idea what port of git I'm using. | 15:40.58 |
henrys | git version 1.7.10.2 (Apple Git-33) - who knows what Apple did to it. | 15:41.06 |
Robin_Watts | I suspect kens ran through the logical order of: Try to install source, try to build it. | 15:41.27 |
kens | will stick with the tools I know I think | 15:41.33 |
Robin_Watts | which meant he'd have hit the need for git before the need for xcode. | 15:41.40 |
kens | I actually installed XCode first | 15:42.07 |
Robin_Watts | and I probably told him to get it from macports, not realising that that wasn't where I'd got it from. | 15:42.14 |
kens | But didn't read anythign to do with it | 15:42.15 |
henrys | but you probably didn't install command line tools from xcode? | 15:44.01 |
kens | Yes I installed those too | 15:44.11 |
| IIRC Xcode hinted that I might need them | 15:44.22 |
henrys | weird a mystery | 15:44.28 |
| you probably just have 2 gits on your system. | 15:44.40 |
kens | Could be :-) | 15:44.47 |
henrys | Robin_Watts, tor8, paulgardiner:I now own a nexus 7 | 15:45.43 |
tor8 | henrys: congratulations, or sympathies? ;) | 15:46.02 |
paulgardiner | Aw shucks! I want to own a nexus 7! | 15:46.37 |
henrys | I guess they are pretty popular here - I picked one up at the local "office" store "Office Depot" | 15:46.38 |
paulgardiner | shivers as he realises that henrys will be hoping for a forms-capable MuPDF app on said nexus 7! | 15:47.56 |
Robin_Watts | paulgardiner: Had any luck with the SEGVing ?> | 15:48.35 |
paulgardiner | Robin_Watts: yes actually :-) Seems ok now. | 15:48.53 |
Robin_Watts | excellent! | 15:49.01 |
| Any idea what it was? OR shouldn't I ask ? | 15:49.10 |
paulgardiner | Robin_Watts: just putting something in the report about it... which isn't to say I really know what it was! | 15:50.12 |
henrys | paulgardiner:yeah that was my next question ;-) | 15:50.58 |
paulgardiner | Progress at last, but still a long way to go, I fear. | 15:51.34 |
| Just sent the report so you can see where we are. | 15:53.00 |
henrys | thanks | 15:53.46 |
sebras | Robin_Watts: sorry for the sebrasism... you fixed it, right? oh and yeah, I meant that it ought to be either axial everywhere or l_or_r... | 15:56.36 |
Robin_Watts | sebras: not yet/ | 15:57.02 |
sebras | Robin_Watts: ok. I'm counting on that you will before it gets merged to master. | 15:59.17 |
Robin_Watts | have just done so locally. | 16:01.37 |
henrys | paulgardiner:wow sounds like you've had an adventure. | 16:14.48 |
paulgardiner | henrys: yes, just a bit. | 16:15.20 |
mvrhel_laptop | good morning | 16:37.33 |
Robin_Watts | Morning | 16:37.41 |
| Updated patches pushed. | 16:43.27 |
marcosw | I just spoke with miles, he can upgrade his internets connection with double the speed for only $5/month. The issue is he has to change from DSL to U-Verse and "bridge mode" won't be available anymore. Does anyone understand what that means? | 17:22.04 |
henrys | yes it means he should get cable | 17:22.50 |
ray_laptop | henrys: if it's anything like my office, cable is REALLY pricey. I pay 3 times as much as I did for DSL | 17:23.54 |
| that takes me from 256Kb upload speed to 768Kb, so I guess it's fair | 17:24.37 |
henrys | bridge mode would mean the router would only work with his computer as far as I know so is not workable. Printers, joanne are on the local net yes? | 17:25.31 |
Robin_Watts | marcosw: You can get 'bonded' DSL, which is where you have 2 DSL lines and run them in parallel for twice the speed. | 17:25.51 |
| It's possible that that is what they mean by 'bridge mode' ? | 17:26.04 |
ray_laptop | Robin_Watts: I think that's what they call it | 17:26.25 |
Robin_Watts | OK,so if you swap to cable, you can no longer pair your connections to get twice the speed. Who cares when cable is much faster anyway? | 17:26.59 |
ray_laptop | Robin_Watts: well, I guess bonding is different (at least according to this guy): http://forums.whirlpool.net.au/archive/367347 | 17:27.51 |
Robin_Watts | Ah. U-verse is FTTC. | 17:28.05 |
| So it's cable to a point closer to your house than the exchange. | 17:28.20 |
henrys | ray_laptop:so is this "business class" cable that is expensive or is it just high in your area? | 17:28.48 |
Robin_Watts | and then hopefully you get faster ADSL over the last few metres than you would going to the exchange. | 17:28.49 |
| So he's never going to get more than ADSL 2+, I guess? | 17:29.08 |
| That's a theoretical maximum of 24Mbps down. | 17:29.28 |
ray_laptop | henrys: it's "business class" that is the issue. I get 20Mb down 2Mb up at home for $39 but in the office for 7Mb down 768K up I pay $160 (or Miles does). And the office is 50 ft from houses that get the cheap rate. | 17:32.07 |
Robin_Watts | marcosw: So Mr Google tells me that "bridge mode" with ADSL is when you tell the router to be dumb, and you let your PC control the picking of the IP address, negotiation with the ISP etc. Miles won't ever be using that. | 17:32.59 |
marcosw | Robin_Watts: see also http://pctech.invisibill.net/?page_id=139 | 17:33.20 |
henrys | ray_laptop:sounds like you need a 50 ft cable | 17:33.24 |
| and a splitter but the cable guys get uptight about that ;-) | 17:35.14 |
ray_laptop | henrys: I considered asking the person in the office next to me if I could put an antenna on their patio, then use Wireless-N to my house, but I think there are too many trees in the way. | 17:35.40 |
| henrys: If I use an ethernet cable from their router, I'd be fine :-) | 17:36.17 |
henrys | so is now a good time to install mupdf on the nexus or is something spectacular going to happen in the next few days? | 17:44.18 |
Robin_Watts | henrys: It won't hurt for you to install it. | 17:45.30 |
henrys | tor8 was saying something about ice cream sandwich but it looks like I have jelly bean. | 17:50.28 |
mvrhel_laptop | bbiaw | 17:50.38 |
Robin_Watts | henrys: Jelly Bean is newer. | 17:50.53 |
| (Jelly Bean = 4.1, ICS = 4.0) | 17:51.18 |
henrys | oh okay - first letter thing like ubuntu | 17:51.28 |
Robin_Watts | Donut, Eclair, Frojo... I guess so. Never realised. | 17:52.10 |
sebras | Robin_Watts: I think there might be an error in patch b652eef1:at the end of the DCT branch of build_filter() you change the default value from -1 to 0... is this intentional? | 19:44.56 |
| also that patch would have been much nicer if you separated out all the renaming stuff (eob, bi1, etc.) into a separate patch. | 19:45.30 |
| would have made it easier to review if nothing else. :) | 19:45.38 |
| also the fz_free_buffer_compressed() -> fz_free_compressed_buffer() seems like it was supposed to be merged into the patch introducing fz_compressed_buffer..? | 19:51.10 |
Robin_Watts | sebras: That was originally part of the monolithic pdfwrite commit. | 20:02.51 |
| and I split it out to make it easier. | 20:03.00 |
| so, I could have split it even further, but it would have been much harder work. | 20:03.24 |
| sebras: New version pushed. Thanks. | 20:07.27 |
| re: compressed_buffer -> buffer_compressed. Yeah. That looks like a cock up. will fix tomorrow. | 20:09.15 |
mvrhel_laptop | this clean up of the rendering intent stuff is sorely needed. this will make life so much easier for me as new stuff needs to be added | 21:02.58 |
malc_ | tor8: hi, around? | 21:39.31 |
aleray | hi, can I encode to base 16 instead of outputting to a file ? | 21:57.51 |
tor8 | malc_: hey | 21:59.55 |
malc_ | tor8: using mupdf-thirdparty-xxx.zip no longer works (without manual tweaking) after submodules were introduced (Makethird expects "shortened" directory names i suppose (haven't looked)) | 22:01.17 |
tor8 | malc_: the unix build system should still work, oh wait, right, it doesn't unless you have directory names without the version numbers | 22:02.19 |
malc_ | aue | 22:03.25 |
tor8 | malc_: hang on, I take that back. the wildcards should still work⦠what errors are you seeing? | 22:03.25 |
malc_ | aye i mean | 22:03.29 |
| i'm not seeing ANY errors, it just doesn't build the thirdparty modules and uses whatever i have installed on my system | 22:03.57 |
tor8 | ah, no, I'm looking at the wrong commit. the thirdparty stuff won't work with the zip file anymore. | 22:04.15 |
malc_ | that was my original point :) | 22:04.29 |
tor8 | malc_: indeed :) any reason you're not using git submodules? | 22:04.42 |
malc_ | tor8: it's for a buildscript (which i _personally_ do not use) that just automates the building of my stuff for less "sophisticated" individuals | 22:05.23 |
tor8 | "git submodule update --init" should do the trick, the thirdparty gits are all hosted at the same | 22:05.34 |
| location as mupdf.git | 22:05.40 |
malc_ | i know where it is located, i know how to use it. that's not really the point, the point is to make this (http://repo.or.cz/w/llpp.git/blob/3d0301835d655aa739281975081ac02333b71ad0:/buildall.sh) work, after all it used to, one way to go about it, is to just update the .zip file with "proper" directory hierarchy | 22:07.19 |
tor8 | malc_: right. I wonder if you can get the gitweb tarball snapshots to do a recursive clone | 22:08.42 |
| or you can download the thirdparty gits in the right place from that script, like you do with the mupdf source | 22:09.18 |
| I want to phase out the mupdf-thirdparty zip; for the next release I was planning to include the thirdparty sources in the main source tarball | 22:10.03 |
| but that's not a final decision, I need to convince my coworkers that it's a good idea first | 22:10.53 |
malc_ | tor8: i don't really care one way or the other (lacking any say in the matter helps alot in being indifferent), i just reckon that even users of vanilla mupdf without git would have trouble building the thing (as god/tor intended) | 22:14.32 |
aleray | how can I convert eps to svg? I don't see svg in the writer list | 22:20.52 |
| (but read somewhere that gs could do it) | 22:21.03 |
henrys | aleray the svg device is unstable and has to be added manually (windows) or autogen.sh (unix) | 22:24.14 |
aleray | henrys, ok, is there any alternative? I'm writting a small web server in python | 22:24.50 |
| (so python bindings would be welcomed) | 22:25.00 |
henrys | I don't know of any alternative but it is not something I've looked around for either. | 22:28.25 |
aleray | henrys, thanks | 22:41.13 |
| do any one know th epython bindings for python? I'l trying to assign the ouput content to a variable rather than saving it to a file | 22:41.51 |
| but I can't capture it in my script and everythong gets printed to stdout | 22:42.14 |
| Forward 1 day (to 2012/10/02)>>> | |