| <<<Back 1 day (to 2015/07/08) | 20150709 |
satish | Hi All, | 09:46.05 |
| Need a small info, when rendering pdf pages in android through mupdf | 09:46.38 |
| Hi, Is anyone there in this IRC? | 09:50.51 |
kens | If you have a question just ask it | 09:51.17 |
| Robin_Watts : ping | 10:31.00 |
Robin_Watts | pong | 11:14.31 |
kens | I'm looking fo advcie on merging my branch with master | 11:14.49 |
Robin_Watts | kens: ok... | 11:15.13 |
licz | Are MuPDF devs here? | 11:15.24 |
kens | Need to merge it an create a single commit, and I'm not too sue how to do that. | 11:15.37 |
Robin_Watts | licz: Yes. | 11:15.40 |
| kens: git checkout master | 11:15.48 |
licz | Is there a way to view 2 pages in mupdf? | 11:15.55 |
Robin_Watts | kens: oh, sorry. | 11:16.03 |
| I'd do... | 11:16.06 |
kens | :-D | 11:16.11 |
Robin_Watts | Have you rebased on master recently? | 11:16.21 |
kens | It shoudl eb bang up to date, I rebased this morning | 11:16.34 |
Robin_Watts | OK, so; git checkout dev_subclassing | 11:16.49 |
| git branch dev_subclassing_squashed | 11:16.57 |
| git checkout dev_subclassing_squashed | 11:17.08 |
| git rebase -i master | 11:17.17 |
kens | OK | 11:17.31 |
| Thenj squash it all flat | 11:17.41 |
Robin_Watts | That will make a new dev_classing_squashed branch, and then change you onto it, and start the squashing process, yes. | 11:17.51 |
kens | OK that'll take me a few minutes | 11:18.10 |
Robin_Watts | licz: There are 2 parts to mupdf. | 11:19.07 |
| There is the main C library core that handles opening/rendering/manipulating PDFs/XPSs etc. | 11:19.35 |
| and there are some example tools built around that, like viewers for various platforms. | 11:19.48 |
| The C core certainly supports rendering many pages at once. | 11:20.05 |
| but I'd guess that you're talking about the android app? | 11:20.25 |
licz | Robin_Watts: no, I'm talking about linux version | 11:22.34 |
Robin_Watts | licz: ok. | 11:22.49 |
| So, the mupdf viewer app on linux only does 1 page at a time. | 11:23.06 |
| If you want something more powerful, look at gsview.com - that's based on mupdf/ghostscript. | 11:23.29 |
licz | ok | 11:24.07 |
satish | For Android PDF viewer can only render one page at a time!! How can i make it multi-pages rendered! | 11:46.19 |
Robin_Watts | satish: By changing the source code. | 11:46.41 |
satish | I meant, which class, do i need to change? | 11:48.01 |
| Basically, am an Android Developer, not so good at C, but will try if someone put a light on it :) | 11:48.29 |
Robin_Watts | satish: The C core does not need to change. | 11:48.51 |
| It is possible that the JNI code might need to be tweaked a bit. | 11:49.05 |
| but most of the work will be in the java files. | 11:49.16 |
| Have you read platform/android/ClassStructure.txt ? | 11:49.37 |
satish | Great Robin! But, using JNI, the app was crashing | 11:50.20 |
| Yeah, I do | 11:50.25 |
Robin_Watts | "using JNI, the app was crashing". You'll need to be a lot more explicit than that. | 11:50.56 |
kens | OK Robin_Watts I've squashed up all the commits and written an (I hope) reasonably clear, if concise, description. How do I get this onto master ? | 11:51.51 |
Robin_Watts | kens: That commit should be sat on the tip of master already, right? | 11:52.16 |
kens | Its on my branch | 11:52.27 |
| which starts at the tip of master yes | 11:52.38 |
Robin_Watts | And the branch is called dev_subclassing2 ? | 11:52.49 |
| And the branch is called dev_subclassing_squashed ? | 11:52.55 |
kens | dev_subclassing_squashed | 11:52.57 |
Robin_Watts | OK: git checkout master && git reset --hard dev_subclassing_squashed | 11:53.14 |
kens | sits on master which asysa 'Bug 696041....' | 11:53.15 |
Robin_Watts | That says: "swap onto master, then make master point to the same thing that dev_subclassing_squashed does" | 11:53.39 |
kens | Hmm woudl never have thougt of that, OK done | 11:53.55 |
| HEAD is now at f98b0b8 Addioitn of device 'subclassin' | 11:54.23 |
Robin_Watts | Alternatively you could have done 'git checkout master && git cherry-pick dev_subclassing_squashed | 11:54.41 |
kens | That was what I would probably have done. | 11:54.53 |
| But ths seems faster | 11:55.00 |
| So, just push to origin now I guess ? | 11:55.14 |
Robin_Watts | yes. | 11:55.45 |
kens | Well, that went quickly O.O | 11:56.12 |
| The cluster is testing it, so I guess its all done, thanks Robin! | 11:57.05 |
satish | Robin: It seems like Mupdf Core is single threaded and only one core can be open at a time. If I try to open multiple cores at the same time, the app crashes. To be more clear, I have an android app that downloads multiple pdf files from multiple urls and displays them. Now, I can wait until all the pdfs are downloaded, concat them and then display but then I need to display my pdf right after the first file gets downloaded . So | 11:57.09 |
| Robin: *but with multiple files from different urls | 11:58.11 |
kens | runs off for some lunch, I'll examine the wreckage later :-) | 11:58.23 |
Robin_Watts | kens: Have you pushed dev_subclassing to golden too? | 12:00.27 |
| satish: MuPDFCore is highly specific to the way our app works. | 12:01.11 |
| It is not a generic reflection of the C interface into java. | 12:01.22 |
| Each MuPDFCore handles just one document. | 12:02.06 |
| But it's certainly possible to have multiple cores, I believe. | 12:02.40 |
satish | Right, I get that core handles a single document at a time but I see that you released progressive download from version 1.3. In progressive download, you get a single file from a server but I have multiple files | 12:04.18 |
Robin_Watts | Yeah, I'm sure we've done a variant that made n 1-page files look like 1 n-page file | 12:05.00 |
| satish: What version of the source are you using? | 12:05.12 |
satish | Really? If so, how? | 12:05.25 |
Robin_Watts | It's not part of the standard source; paulgardiner did it for a customer, I think. | 12:06.00 |
| satish: What is the aim of this stuff? Are you writing an app for general release? | 12:06.26 |
| You're aware of the MuPDF licensing situation, right? | 12:06.39 |
| MuPDF is released under 2 licenses. The first is the GNU AGPL. This is a complex license, but simplisitically, it says that you can use MuPDF in your own project, on the condition that anyone you distribute your project to must be able to get a copy of the source code to the *entire* project on request. | 12:07.56 |
satish | Robin: Yes, it is for general release. | 12:07.56 |
Robin_Watts | Hence most people writing an app can't go along with that (as it would mean giving away their entire source code). | 12:08.27 |
| The other license is the Artifex Commercial License. This is where you sign an agreement with Artifex, and pay us a fee, and you get to use the code within your own app as you see fit. | 12:09.16 |
| This may or may not include support from us (depending on the exact deal you sign). | 12:09.31 |
satish | Robin: Here is a question i have: if i have a pdf page, then i create a core and call drawpage. Now, this page is drawn on my screen. While i am reading this page, can i render my other pages/files in the background simultaneously? | 12:10.24 |
| What is the other license available? | 12:10.33 |
Robin_Watts | The other license is the Artifex Commercial License. This is where you sign an agreement with Artifex, and pay us a fee, and you get to use the code within your own app as you see fit. This may or may not include support from us (depending on the exact deal you sign). | 12:11.05 |
satish | Okay, got it! | 12:11.11 |
Robin_Watts | satish: I can't think why not. | 12:11.48 |
satish | Robin: I am planning to make my project open source | 12:12.23 |
| Robin: what is the function (native) we need to call to render a file in background. only render and keep in memory and not draw on screen | 12:13.18 |
Robin_Watts | It would need to be GNU AGPL. | 12:13.24 |
satish | I was just pondering at that idea of AGPL vs GPL v2 | 12:15.04 |
Robin_Watts | satish: Either you need to work within the existing apps structure (in which case everything goes through MuPDFCore) or you need to build it from the ground up. | 12:15.48 |
satish | Robin: I am not a C programmer. I am willing to modify the java code to make it work | 12:17.08 |
| Robin: If i would like to have mupdf customized to my needs by you on an agreement base, how much would that cost me? | 12:18.56 |
| Robin: Also, I am using mupdf version 1.2 right now | 12:24.43 |
kens | Robin_Watts : I didn't push the squashed up branch, its the same as the existing device_subclassing branch except that it has no history in it. | 12:24.43 |
Robin_Watts | kens: Yeah, it was the unsquashed one I was asking about. | 12:25.23 |
| satish: I can't comment on costs. That's not my area. | 12:25.34 |
| Early versions of mupdf had problems with multiple instances. | 12:25.50 |
kens | Robin_Watts : the unsquashed one's been there for a while, its what Marcos has been doing regression tests with | 12:26.00 |
satish | Robin: Could you direct me to the right person? Thanks! | 12:26.06 |
Robin_Watts | 1.2 is 4 years old-ish. | 12:26.15 |
| We're on 1.7a now. | 12:26.20 |
| 3 years old. | 12:26.38 |
satish | Robin: Which version is safe to opening multiple cores? | 12:27.12 |
Robin_Watts | You should definitely retry with the latest version - it might solve your problems. | 12:27.14 |
| satish: I can't remember without digging through the history. | 12:27.27 |
| You'd need to contact sales@artifex.com and talk to Scott about getting NRE done. | 12:28.03 |
| The problem is, we're very busy at the moment. | 12:28.14 |
satish | Robin: Let me get this clear then. If i use 1.7 say, can i open multiple cores at the same time and navigate from one core to another? | 12:30.21 |
Robin_Watts | I would certainly think so. | 12:30.32 |
| Well, let me rephrase... | 12:30.46 |
| I believe that using 1.7 you should be able to open multiple cores at the same time. | 12:31.18 |
| The java classes will need to be updated to use a different core per page, but the C should not need to change. | 12:31.44 |
| Really, use 1.7a not 1.7. | 12:31.53 |
| or better, the version from git. | 12:32.01 |
satish | Robin: if i open multiple cores at the same time and here is the scenario: first page path goes to core1 and second to core2. Now both cores are open at the same time. will the pages be rendered and ready when i swich from one core to another or will it show that loading spinner again? | 12:32.53 |
Robin_Watts | We cache the rendered pages in the core. | 12:33.37 |
| Hold on, let me confer with a colleague. | 12:34.03 |
satish | I have already changed my java code to match this. this is what i did: made a new core with page one loaded. the user swipes, then i call a new core again with page 2 path. but the problem with this is everytime i swipe for a new page/core, it always shows or takes time to load first with the loading spinner | 12:35.01 |
Robin_Watts | satish: If you create a new core each time, then yes, it will load each time. | 12:39.39 |
| OK, so my colleague wrote some code for this ages ago. It can found here: | 12:39.59 |
| http://git.ghostscript.com/?p=user/paulg/mupdf.git;a=commitdiff;h=4720a73ecb3eb62c3912dffd3520cb54e9157cc9 | 12:40.49 |
| This is NOT ideal. | 12:41.03 |
satish | Robin: So how can i navigate between multiple files using a single core and load it all only once? | 12:41.06 |
Robin_Watts | satish: You can't, as the code is currently written. | 12:41.23 |
| At the moment, the MuPDFCore stuff assumes 1-core = 1 file. | 12:41.35 |
| You'd have to make changes inside the JNI code to change that assumption. | 12:41.59 |
| What paul has done in his branch is to change the java so that it keeps multiple cores open, and moves between them for pages. | 12:42.26 |
| This is likely to be less than ideal for memory usage. | 12:42.44 |
| Within a core, we arrange to only cache the current page and the page above and below it. | 12:43.07 |
| With Pauls way of working, you'd end up potentially with all pages cached, and you could easily have memory problems because of that. | 12:43.33 |
satish | You think i could just keep the previous 2 and the next two files in cache using Paul's code? | 12:44.51 |
Robin_Watts | satish: I think you'd need to make changes at the JNI level to do that. | 12:46.12 |
| It might be possible - I haven't looked at it in detail. | 12:46.24 |
satish | Robin: Could you give me a hint on what kind of changes i can make? Sorry, I am not a pro developer. | 12:47.19 |
| Thanks Robin for your time. I guess I will try Paul's code and I will contact you if I have further questions. | 12:55.18 |
Robin_Watts | satish: go for it. I am limited in how much time I can spend on this... sorry. | 12:55.42 |
sebras | kens: what did you push? master? | 12:58.18 |
kens | A massive branch to master | 12:58.41 |
| THs is Ghostscript though | 12:58.56 |
sebras | kens: ah. | 12:59.34 |
kens | cchrisl I just ran across a problem with the strtok stuff, GhostPCL won't build on VS because there's a variable declaration inside the procedure body. Want me to fix it ? | 13:55.32 |
| chrisl^^ | 13:55.49 |
chrisl | kens: I have it fixed locally - I am confused by that..... | 13:56.01 |
kens | Ah OK, I only found it becase I was fixing a potential problem in my code :-) | 13:56.32 |
chrisl | kens: I don't understand how there wasn't an error before on that.... | 14:00.01 |
kens | Hmm maybe gcc doesn't care ? | 14:00.16 |
chrisl | It was gcc that threw it here | 14:00.30 |
kens | O.O | 14:00.34 |
| I'm baffled then | 14:00.54 |
chrisl | IIRC, I have the option set to make it an error in gcc, so the cluster should catch it | 14:01.11 |
kens | I have no clue, it certainly was working before. I'd blame me and my branch commit, but I can't see how I'd end up with part of a fix. | 14:01.49 |
| No, the original commit has it ther | 14:02.18 |
| Clearly the cluster doesn't throw that error | 14:02.38 |
chrisl | I also built it on Windows, so...... | 14:02.53 |
| I'm just cluster testing now | 14:03.22 |
kens | me toot :) | 14:03.34 |
chrisl | Oh, well, I'll kill mine, and let you commit it | 14:04.09 |
kens | Not a big deal | 14:04.28 |
| I needed to fix it to test my own change | 14:04.42 |
rayjj | morning, floks | 14:06.23 |
kens | sigh, should have made it a pcl only test, oops | 14:06.24 |
rayjj | folks | 14:06.25 |
kens | Mornign rayjj | 14:06.28 |
rayjj | I like that. Corresponding typos :-) | 14:06.48 |
| kens: I saw that you pushed the device_subclassing | 14:07.25 |
kens | Yep, nobody objected | 14:07.39 |
rayjj | Now I need to try and see if my repo will update and merge cleanly (for my fast HT wip) | 14:08.11 |
kens | It *should* be OK...... | 14:08.25 |
rayjj | kens: no objection! Congratulations !!! | 14:08.32 |
kens | Thanks :-) | 14:08.38 |
| Remind me never to do anythign like that again...... | 14:08.46 |
rayjj | kens: OK. | 14:08.51 |
| remind me never to muck around in the image code again. Particularly when MV and RJW have been in there since the original (that no one remembers why it was done certain ways) | 14:10.15 |
kens | That all sounds prefectly normal for Ghostscript..... | 14:10.44 |
rayjj | I've determined that the mono and color cases use different, non-compatible, pixel stepping rules. I had fixed the mono case, then worked on the color case and discovered that what fixed that broke the mono case :-( | 14:11.49 |
| that and some places take penum->adjust into account and some do not, but AFAICT penum->adjust is always 0 for images | 14:13.26 |
kens | Then what's hte point of it being in the image enumerator ? Or am I missing something ? | 14:14.44 |
rayjj | I assume it is so we can use different 'adust' rules for images and fills | 14:16.20 |
| generally, for PS image rules, pixels are supposed to paint from the image IFF the source pixel covers the center of a device pixel, but fill paints device pixels if the source shape touches any part of it | 14:17.55 |
| kens: which I'm sure you know | 14:18.10 |
| sorry for re-stating the obvious | 14:18.25 |
kens | rayjj we don't use an enumerator for fills, and if the image one is always 0, then why bother with it in the image enumerator ? | 14:25.22 |
| Seems poitnless, but I guess its probably another relic. | 14:25.41 |
Robin_Watts | kens, rayjj: Force it to 0, cluster test ? | 14:26.12 |
kens | Cluster testing is good but not perfect, it might be non-zero for some weird device or something | 14:27.16 |
Robin_Watts | yes, it's imperfect, but it's a cheap test that may at least show that it *is* used. | 14:27.42 |
| (I bet it will show that it's used) | 14:31.59 |
kens | I'm not sure its practical to force it to0, you;d have to identify where each enumerator is created and each occasion where teh adjust value is set. | 14:35.36 |
Robin_Watts | kens: If you can see where it's used, you can either set it to 0 there, or nobble it so that it's always treated as 0. | 14:36.23 |
| just an idea. | 14:36.26 |
kens | I just thnk that there might be quite a few places, and that kind of review could probably tell you whether its ever non-zero anyway | 14:36.51 |
| But its up to ray | 14:36.59 |
rayjj | kens: Robin_Watts: We'd need to force it non-zero to see where it is used. AFAICT it is always 0 now. But I know that it is used in some cases and not in others. In fact, one place where it is *not* used has a comment about that by RJW (whoever that is) | 15:01.11 |
Robin_Watts | I'd not trust anything that RJW bloke says. Very unreliable. | 15:05.05 |
| rayjj: I reckon you're wrong though. We want to force it to zero, so that the cluster will show differences whenever it used to be non-zero. | 15:06.26 |
kens | thinks try t both ways | 15:07.22 |
rayjj | Robin_Watts: oh, so you think there are cases where it is currently non-zero. I misunderstood | 15:07.23 |
Robin_Watts | rayjj: Naively, you'd expect adjust to always be 0 for images. | 15:07.54 |
kens | It appears to be potentially set to non-zero for image masks | 15:08.20 |
| gximage1.c, gx_begin_image | 15:08.34 |
Robin_Watts | Therefore for the code to be in there to cope with it being non-zero, you'd expect that it's set non-zero *somewhere*. | 15:08.40 |
| kens: Ah. | 15:08.43 |
kens | penum->adjust = | 15:08.46 |
| (pim->ImageMask && pim->adjust ? float2fixed(0.25) : fixed_0); | 15:08.46 |
Robin_Watts | kens: Urgh. That's a hack then. | 15:08.57 |
kens | Yeah I'd say it is, though what its for..... | 15:09.13 |
Robin_Watts | It's for "perforamance", apparently. | 15:11.09 |
kens | Oh, where did you find that ? | 15:11.18 |
Robin_Watts | $ git log -1 54babf72 | 15:11.30 |
| commit 54babf720cbab2d495dc7730a5d97824efbfcc68 | 15:11.32 |
| Author: Henry Stiles <henry.stiles@artifex.com> | 15:11.34 |
| Date: Fri Oct 2 08:45:22 1998 +0000 | 15:11.36 |
| perforamance | 15:11.38 |
| git-svn-id: http://svn.ghostscript.com/ghostpcl/trunk/ghostpcl@453 06663e23- | 15:11.39 |
kens | :-) Then ray can beat up Henry about it.... | 15:11.52 |
Robin_Watts | (Apologies to Henrys, this must be like when parents bring out baby photos) | 15:12.03 |
henrys | good think I'm on vacation today | 15:12.24 |
kens | Crumbs there's a lot of changes in ther for such a short commit log | 15:12.54 |
rayjj | kens: not as many as "DeviceN" ;-) | 15:13.16 |
Robin_Watts | henrys: Whatever you do, don't speak up and tell people you're here. | 15:13.25 |
henrys | fixing ... | 15:14.01 |
marcosw_ | just to confirm, there werenât any changes to the encryption/decryption code between 9.15 and 9.17? I looked at the logs and didnât see anything. | 15:14.02 |
kens | I dont' recall a | 15:14.35 |
| changing anything | 15:14.39 |
| adjust | 15:14.51 |
chrisl | marcosw_: quick cluster question: does the cluster do "make install" on ghostscript (and/or pcl and/or xps)? | 15:15.13 |
kens | Hmmm gximage.h had a 'adjust' before that commit of Henry's | 15:15.39 |
marcosw_ | yes on ghostscript, let me check on the ghostpcl/xps | 15:16.26 |
Robin_Watts | kens: I was referring to the "0.25" value being used for adjust. That's a hack. | 15:19.15 |
marcosw_ | yeah, no make install for ghostpcl/xps, just copies the binary. | 15:19.16 |
kens | Robin_Watts : oh, yes possibly, but it must have had *some* meaning before | 15:19.36 |
chrisl | marcosw_: okay, thanks. With the new build stuff, "make install" will install all the executables..... | 15:19.55 |
Robin_Watts | kens: 0 and 0.5 (or 0.5-epsilon) are 'expected' values for adjust. | 15:20.09 |
kens | Ah, OK then | 15:20.17 |
marcosw_ | chrisl: that sounds like it will require changes to the cluster code :-) | 15:20.58 |
kens | Oh great, a 68 page example file..... | 15:22.13 |
chrisl | marcosw_: this will definitely need changes to the cluster code - directories are "flatter", only one configure script, only one make call, all the executables end up in the same "bin" directory..... | 15:22.18 |
rayjj | chrisl: but still separate obj dirs, correct ? | 15:22.53 |
marcosw_ | kens: I tried to simpllify it but Acrobat give an error when I try to delete pages. Something about the file structure being invalid (I decided not to mention that in the bug report :-) ). | 15:23.36 |
rayjj | or did you mean all of the parser app binaries (gs, gxps, pcl6, ...) in the same bin (bin, debugbin, profbin, ...) | 15:23.50 |
kens | marcosw, its primarily the reason we can't renmder it, the PDF is broken | 15:23.55 |
chrisl | rayjj: no, shared obj dirs, too | 15:24.07 |
rayjj | chrisl: shared between 'type' of build (profile, debug, release, ...) ???? | 15:24.34 |
chrisl | Yes, same as Ghostscript works now | 15:24.44 |
kens | marcosw looks like WPCubed broke the file to me. | 15:24.45 |
marcosw_ | chrisl: the cluster wants all the binaries in the same diretorry (~/cluster/gs/bin). Thatâs used as âprefix with autogen.sh and itâs where the pcl/xps binaries are copied to. | 15:24.56 |
rayjj | chrisl: gs has debugobj and debugbin now | 15:24.59 |
Robin_Watts | chrisl: I hope that's a "no, same as gs works now" :) | 15:25.02 |
chrisl | I took ray's comment to mean all "profile" builds share, all "debug" builds share etc | 15:25.56 |
| So "make debug" still ends up with "debugobj" and "debugbin" | 15:26.32 |
rayjj | chrisl: no, Robin_Watts had my meaning. So what you have is the pcl6, ps, xps, ... objs all together in obj, debugobj, profobj, ... | 15:27.01 |
Robin_Watts | That sounds perfectly sane, thanks for clarifying. | 15:27.03 |
rayjj | chrisl: sounds ok to me | 15:27.16 |
marcosw_ | kens: of course the file is broken, I think 100% of the reported files from that customer are broken. The problem is that all the other software they try opens the file correctly (and without warning), so their thinking is thatâs it broken in a way we should be able to handle. | 15:27.26 |
chrisl | Yes, sorry - my mistake | 15:27.26 |
kens | marcosw I'm not sure that's going to be possible, if you try mupdfclean on it for example it produces a broken PDF file that Acrobat won't open. | 15:28.00 |
marcosw_ | chrisl: is the new âflatterâ build system available in a public branch? Iâd like to make the cluster changes before you commit it to master. | 15:29.03 |
rayjj | kens: so open a bug against mutool clean ;-) | 15:29.07 |
chrisl | marcosw_: It's on my repo just now. I have a couple of tweaks to the Windows build, then I'll push the branch to golden | 15:29.43 |
Robin_Watts | golden? Why golden? | 15:30.05 |
rayjj | marcosw: so your plan is to have the cluster cope with old and new structure, right ? | 15:30.21 |
marcosw_ | kens: if I tell the customer that mupdfclean breaks the file heâll point out: âthen donât run mupdfclean on itâ (a reference to the old joke, âdoctor, it hurts when I do thisâ). | 15:30.22 |
| rayjj: thatâs the goal :-) | 15:30.39 |
kens | marcosw I know, I'm just pointing out that we cannot guarantee to render any given broken file,m just becasue 'Acrobat does it'. There are files we render and Acrobat does not for instance | 15:31.03 |
marcosw_ | rayjj: if itâs possible will depend on how many things chrisl changed. | 15:31.17 |
chrisl | Robin_Watts: what else should I call it? | 15:31.19 |
kens | THe object 290 0 in the xref poitns to an object whose index number is actually 222, that';s part of theproblem | 15:32.00 |
Robin_Watts | I would have expected your branch only to go to golden once it had been completely tested, and was about to be squashed/merged to master. | 15:32.01 |
marcosw_ | kens: donât mention that we render files that acrobat doesnât! Customers will start reporting that as bugs (i.e. we use Ghostscript to preflight before rendering with Acrobat, so we need both to fail). | 15:32.10 |
kens | I can make files fail with no problem at all :-) | 15:32.31 |
chrisl | Robin_Watts: marcosw_ tests the branch from golden, not from our individual repos | 15:32.33 |
rayjj | my kids are arriving Heathrow on Sunday -- how accurate is the forecast for rain that day ? | 15:32.50 |
Robin_Watts | rayjj: Hold on, I have a coin I can flip somewhere... | 15:33.04 |
marcosw_ | thinks maybe heâll fire ua a couple of aws instances to use as a test cluster. | 15:33.23 |
| ^ua^up | 15:33.31 |
chrisl | Isn't Sunday Wimbledon finals day? Guaranteed to rain..... | 15:33.55 |
Robin_Watts | rayjj: My phone says 50% chance of rain here on sunday. | 15:34.00 |
rayjj | Robin_Watts: It's just that a news program here was saying how much better UK and European weather forecasting was compared to the US | 15:34.08 |
| chrisl: Oh, that nails it. Thanks :-) | 15:34.28 |
kens | marcosw well there are two pages with the same object number, I can't see any way that's ever goinhg to work. Whch one should we use ? | 15:34.54 |
Robin_Watts | kens: The right one, obviously! | 15:35.08 |
rayjj | and if you want to be doubly sure, wash the car that AM and leave it parked outside | 15:35.12 |
kens | Robin_Watts : yeah well that's the problem, the one we are using is the one with no MediaBox | 15:35.22 |
chrisl | rayjj: I was just typing that! | 15:35.30 |
marcosw_ | I was watching the british GP on Sunday and thought they had moved the race to France or somewhere with reasonable weather since there was actual blue sky for the first hour. Then it started to rain and everything was again right with the world. | 15:35.31 |
Robin_Watts | rayjj: I can sprinkle some lawn food if you want. That guarantees dry weather. | 15:35.59 |
kens | It nwas blazing hot on Saturday | 15:36.07 |
Robin_Watts | marcosw_: I was out running when it started to rain. | 15:36.24 |
| I thought, aha, the race will be interesting :) | 15:36.33 |
marcosw_ | oddly enough we had rain yesterday, but thatâs my fault. I had a kite surfing lesson scheduled. | 15:36.42 |
| Robin_Watts: all formula 1 races should be in the rain (particularly this season, where the only racing is between teamates). | 15:37.14 |
Robin_Watts | marcosw: When did you start smoking dope? :) | 15:37.14 |
rayjj | marcosw: what, you were afraid of getting wet? You must be a really advanced kite surfer ;-) | 15:37.27 |
kens | rayjj : weather forecasting is prety good 24 hours in advance, drops off with time after that. 3 days is about the limit of predictability | 15:37.30 |
marcosw_ | Robin_Watts: in college. but what does that have to do with anything? | 15:37.38 |
| rayjj: it would have been my first lesson. So a lot of it happens out of the water (you learn kite control first). | 15:38.17 |
kens | rayjj surely you weren't going to send your kids here without a raincoat ? | 15:38.24 |
Robin_Watts | I thought the instructions for kite surfing were similar to snowboarding. Step 1) Light spliff, Step 2) unpack equipment,... | 15:38.47 |
kens | marcosw I was talking to some folks in Clacton doing kite surfing lessons, they reckoned it wa day 2 before you got in the water, day 1 was all about kite control | 15:38.59 |
rayjj | kens: no, it's just whether or not it's in their backpack or checked bag | 15:39.08 |
chrisl | marcosw_: FWIW, if you want to pull the branch from my repo, it's called 'build_consolidation'..... | 15:39.14 |
kens | rayjj well they'll enter the terminal so, no big deal, its dry indoors :-) | 15:39.29 |
| Unless you go to Sochi of course | 15:39.38 |
chrisl | has to shuffle off - will check in later....... | 15:40.13 |
marcosw_ | kens: Iâve been flyint two and four line kites for ~20 years, so am expecing that Iâll have an easier time with that part. otoh, I have a terrible sense of balance so not falling over while standing on the board willl be a problem. | 15:41.26 |
kens | marcosw they did say if you were experienced wiht kites they might get ot body dragging on the first day, and if you were good, maybe to standing on the board | 15:41.57 |
| But almost everyone doesn't get that far till day 2 | 15:42.13 |
| They were doign 2-up as well, £5 for 10 minutes | 15:42.36 |
| If I'd been there 2 days I';d have been tempted, it was hot on the beach. Ended up borrowing a kayak and paddling around the pier instead. | 15:43.05 |
marcosw_ | has to run as well. probably back later. | 15:43.18 |
kens | Oh boy, ths is really good *every* page is defined twice in ths file, and the second definition is broken. | 15:43.49 |
| Basically WPCubed f**ked gthe file totally | 15:44.08 |
| Actually I take it back, not *every* page is defined twice, some of them are only defined once, but share their object number with other objects in the file. | 15:45.20 |
Robin_Watts | mvrhel_laptop: ping | 17:13.43 |
mvrhel_laptop | bbiab | 19:02.54 |
| Forward 1 day (to 2015/07/10)>>> | |