| <<<Back 1 day (to 2013/04/03) | 2013/04/04 |
_DomY-Dom | http://pastie.org/pastes/7307298/ | 00:30.19 |
Ovitz | Hi guys, I have question about MuPDF port for WinRT. Do you plan make port of your library for Windows RT? | 07:07.53 |
kens | I'm not aware of any such plans, but I'm not working on MuPDF< so I could be wrong | 07:09.22 |
| tor8 ping | 07:10.29 |
tor8 | you'll have to ask mvrhel about that | 07:10.56 |
| but I believe we did look into it some time ago | 07:11.15 |
kens | I thought Michael was doing a WIndows 8 desktop viewer ? THat won't beRT I don't think | 07:11.17 |
tor8 | Robin_Watts may know more, I wasn't paying attention (hey, it's WinRT...!) | 07:11.43 |
| there were a few minor issues with syscalls and the sandboxing, but nothing too major IIRC | 07:12.08 |
| you could always rip out the file operations and work with files pre-loaded into memory | 07:12.36 |
Ovitz | OK. Do you know who can give me answer on this question? Because I need 100% sure. We are searching pdf library for commercial product. | 07:15.23 |
tor8 | Ovitz: hang around for a few hours and you'll be sure to get an answer. many of us are still asleep. | 07:16.11 |
Ovitz | OK, I will, thanks. | 07:17.16 |
tor8 | Ovitz: http://bugs.ghostscript.com/show_bug.cgi?id=693668 is that from you? | 07:20.25 |
Ovitz | No | 07:24.12 |
kens | chrisl ping | 07:45.02 |
chrisl | kens: pong | 07:46.53 |
kens | DO we still have a way to build GS on WIndows as a single executable, rather than using the DLL ? | 07:47.15 |
chrisl | No, AFAIK, that went away before I came along | 07:47.55 |
kens | Oh well, scratch that then, never mind. | 07:48.06 |
chrisl | kens: actually, there is still a target in msvc.mak for a "big" exe, but I've no idea how to get to it...... | 07:51.12 |
kens | No me neither, which is why I was asking :-) | 07:51.27 |
chrisl | Ah, try adding MAKEDLL=0 to the nmake command line | 07:52.29 |
kens | Oh interesting, just a moment | 07:52.40 |
| just registering my VS 2012 express | 07:52.51 |
| As Bob Howard working, for No Such Agency using the MI5 PO Box | 07:53.18 |
chrisl | :-) | 07:53.42 |
kens | OK clena done, rebuilding now | 07:55.21 |
| clean* | 07:55.24 |
| I want the customer with the thread error to try a non-DLL executable if possible, because I think that is what is causing the problem. | 07:56.00 |
| Well, its some M$ rubbish causing the problem, but this might alleviate it, and prove its not us. | 07:56.23 |
| ah, and now I have a 10Mb executable with no DLL, excellent | 07:57.52 |
| thanks chrisl | 07:57.58 |
chrisl | np - I'm astounded it still worked! | 07:58.11 |
Robin_Watts | Ovitz: hi | 09:14.19 |
| We do indeed have a projects underway to get a mupdf viewer running on Windows RT. | 09:14.44 |
| Michael Vrhel (mvrhel) is the guy doing it. | 09:15.20 |
| He has been working on Windows 8, but it works on ARM based windows devices too. | 09:15.53 |
| His progress can be seen here: http://git.ghostscript.com/?p=user/mvrhel/mupdf.git;a=shortlog;h=refs/heads/winRT | 09:16.44 |
Ovitz | OK, thanks. Do you know when beta version will be ready? | 09:21.10 |
Robin_Watts | Ovitz: You can download it and build it now. | 09:22.16 |
| Michael just has a few things like reflow support to add, AIUI. | 09:22.34 |
Ovitz | Och | 09:38.24 |
Robin_Watts | Interesting. gs_alloc_fill can get called with len being negative. | 09:46.31 |
kens | THat sounds... odd. | 09:46.42 |
Robin_Watts | Anyone here understand the way gs allocates strings ? | 09:56.43 |
kens | chrisl does :-) | 09:56.50 |
Robin_Watts | It's the second call to gs_alloc_fill in i_resize_string | 09:58.19 |
Ovitz | Robin_Watts: I have problem with download source from git | 10:00.40 |
kens | What is the problem ? | 10:00.56 |
Ovitz | I got this message when I try download source via tortoise: fatal: remote error: access denied or repository not exported: /user/mvrhel/mupdf.git | 10:03.05 |
kens | Hmm osunds like Michael hasn't made it public or something | 10:04.19 |
| Robin_Watts : tor8 ? | 10:05.05 |
Robin_Watts | Ovitz: oh. let me try. | 10:05.32 |
| Ovitz: The snapshot links work on the link I gave above. | 10:07.19 |
tor8 | Robin_Watts: I looked through the winRT branch, a few notes (also for mvrhel for the logs): | 10:08.50 |
| there are still #ifdef c++ guards in the header files (ick!) | 10:09.07 |
| one of the commits clobbers the win32 project file (makes it VS2012 not VS2005) | 10:09.28 |
| the winRT directory should be lowercase | 10:09.46 |
Robin_Watts | tor8: We *should* have C++ guards in the header files. | 10:09.48 |
tor8 | Robin_Watts: NO! NEVER! | 10:09.54 |
Robin_Watts | I hate C++ as much as you, but it costs us nothing, and saves other people work. | 10:10.21 |
tor8 | also, I wonder if there's a way to reuse the VS project files that are duplicated between winrt and win32 | 10:10.22 |
Robin_Watts | Ovitz: It should work now. | 10:10.36 |
| But be aware that we may be rewriting the history of this branch before it becomes properly public. | 10:10.53 |
tor8 | it makes us complicit in spreading the disease that is c++ | 10:10.54 |
Robin_Watts | chrisl: ping ? | 10:18.50 |
Ovitz | Robin_Watts: Now it works, thanks. | 10:19.10 |
Robin_Watts | Ovitz: Please let us know how you get on with it. | 10:19.46 |
chrisl | Robin_Watts: pong (sorry, more bike related torture!) | 10:23.19 |
Robin_Watts | chrisl: no worries, I'm about to head out for a run. | 10:23.35 |
| I wondered if you could have a quick look at i_resize_string, and try and understand the second call to gs_alloc_fill. | 10:24.10 |
| it seems to be capable of asking to fill negative numbers of bytes. | 10:24.33 |
chrisl | Robin_Watts: just getting the file up now, I'll have a loko | 10:24.39 |
| or look, even! | 10:24.44 |
Robin_Watts | If you can't see it easily, then don't worry. i'll scribble some diagrams when I get back from running. | 10:25.07 |
| I think it's the HDR_ID_OFFSET stuff that's breaking it. | 10:25.24 |
chrisl | You can check that by disabling it | 10:25.42 |
| Robin_Watts: old_num - new_num can be less that 0 | 10:26.51 |
| Oh, scratch that, looking at the wrong place...... | 10:27.45 |
| Robin_Watts: Yes, I think you're right - the HDR_ID_OFFSET calculations should be removed. I think I forgot that I'd done the pointer offset and length changes in one place at the top of the function. | 10:33.41 |
| I've made the change locally, and I'm running a bunch of tests with -Z@ now | 10:37.21 |
Ovitz | Robin_Watts: It's seems that one file is missing: LVContents.h/cpp | 10:52.09 |
| And I can't compile source | 10:52.26 |
tor8 | Robin_Watts: http://source.mozillaopennews.org/en-US/articles/introducing-tabula/ | 11:06.10 |
Robin_Watts | Ovitz: You probably need to hang around for about 3 or so hours until mvrhel appears. | 11:54.52 |
| tor8: I just so don't want to have to think about another way of doing stuff :) | 11:57.17 |
| tor8: My next job is to try to extract images for reflow. | 11:59.23 |
| Hence I need some way of representing the 'flow' of text/images. | 11:59.40 |
| Either I need to update our 'block/line/span' structures to allow images in there too, or I need to introduce a higher level structure to contain both block/line/spans and also images. | 12:00.43 |
| tor8: Do you have any strong opinions before I start work? | 12:01.13 |
chrisl | Robin_Watts: The test just completed, no crashes or unexpected errors: here's the patch http://pastebin.com/YZRBD8Dy | 12:28.05 |
| Do you want to try it with valgrind? | 12:28.12 |
Robin_Watts | chrisl: I will, ta. | 12:30.47 |
chrisl | bbiab | 12:31.12 |
Robin_Watts | chrisl: That looks good to me. Want to commit it ? | 12:51.28 |
tor8 | Robin_Watts: not really. maybe have a tagged union/superclass thing for block/image, or have two lists at the page level, one of blocks and one of images | 12:56.18 |
Robin_Watts | tor8: I think I'd rather have the page level be a list of {type, ptr} | 12:56.49 |
| So images are at the same level as blocks within the flow of the page. | 12:57.23 |
| That's equivalent to your first suggestion, actually. | 12:57.49 |
tor8 | yes. | 12:58.12 |
Robin_Watts | having 2 lists means it's hard to represent the order of the two classes. | 12:58.18 |
| ok, I shall start on that. | 12:58.24 |
| paulgardiner: ping | 12:58.33 |
paulgardiner | hi | 12:58.40 |
tor8 | did we agree on what to do with the android UI yesterday? | 12:58.48 |
Robin_Watts | paulgardiner: The review you put up yesterday... there seems a fair amount of repeated code. | 12:59.15 |
| Can we not just catch (Exception e) and do it once ? | 12:59.24 |
| tor8: I got the impression that you/paulgardiner had more or less agreed on something, and it sounded reasonable to me. | 13:00.01 |
| I mailed paulgardiner a zipfile of icons last night. Did you get those? | 13:00.21 |
tor8 | two buttons that pop up menus, was it? | 13:00.21 |
| one for annots and one for the kitchen sink | 13:00.29 |
paulgardiner | Robin_Watts: Yeah saw your comment and updated it | 13:00.52 |
tor8 | Robin_Watts: no, I did not get a zip of icons | 13:00.54 |
Robin_Watts | tor8: I didn't copy to you. It was just (+) and (v) | 13:01.11 |
paulgardiner | Robin_Watts: icons look a bit rough because of lack of antialiasing | 13:01.18 |
Robin_Watts | I can forward if you want. | 13:01.19 |
| paulgardiner: They should be antialiased. | 13:01.27 |
paulgardiner | hmmm | 13:01.33 |
Robin_Watts | Are you viewing them in the app ? | 13:01.40 |
paulgardiner | yeah | 13:01.44 |
Robin_Watts | then I probably extracted them wrongly. Let me retry. | 13:01.54 |
| paulgardiner, tor8: new icons sent | 13:17.35 |
paulgardiner | ta | 13:17.46 |
| Robin_Watts, tor8: http://intranet.glidos.net/~paul/MuPDF-new-menu.apk | 14:08.15 |
tor8 | Robin_Watts: I like those icons | 14:08.28 |
paulgardiner | The changes are on paul/master | 14:08.28 |
chrisl | Robin_Watts: yeh, I'll commit that patch, thanks | 14:14.15 |
| Phew, well, after three tries (the second two in the snow!!) I now have an external alarm box for my wireless burglar alarm..... what a chore! | 14:18.17 |
kens | snow finally stopped here | 14:23.16 |
chrisl | It's actually almost stopped here - just ~5 minutes after I put the ladder away! | 14:23.40 |
henrys | chrisl:wow I haven't even locked my doors in years and I just assumed it would be better there. | 14:25.21 |
kens | In the UK ? I think not.... | 14:25.39 |
| maybe in a village | 14:26.06 |
chrisl | henrys: I'm probably being a bit paranoid - but with some fairly expensive stuff around my house, and the possibility of leaving it empty for a few days every now and then, I felt it would be safer. | 14:26.48 |
henrys | although I have friends that now lock up because they've had bears come in ;-) but there a bit west of me and more in the country. | 14:27.35 |
chrisl | Plus, in theory, I should get a small discount on my house contents insurance (although they'll probably change the rules on that!) | 14:27.41 |
kens | Well at least bears we don't have to worry about.... | 14:27.55 |
chrisl | I think the siren I just fitted would scare off a bear - damn near deafened me! | 14:28.53 |
henrys | have a nice day guys! I'm off for the day. | 14:36.30 |
Robin_Watts | henrys: have fun. | 14:36.57 |
| mvrhel_laptop: Morning. | 14:37.17 |
| Ovitz (who just disconnected it seems) has been trying to build your winRT branch and failing. | 14:37.37 |
| Ovitz>Robin_Watts: It's seems that one file is missing: LVContents.h/cpp | 14:38.09 |
| <Ovitz>And I can't compile source | 14:38.11 |
| mvrhel_laptop: Also, tor8 looked over your branch and had some comments. | 14:38.32 |
| See the logs earlier. | 14:38.44 |
| Can't we just rewrite the history of the branch so that you never touch the win32 projects? | 14:39.02 |
mvrhel_laptop | Robin_Watts: yes. I need to do a commit there are several fixes needed | 15:00.57 |
| Robin_Watts: You are probably able to rewrite the history | 15:01.41 |
Robin_Watts | mvrhel_laptop: tor8 also objected to the use of "winRT" as a directory name (he'd prefer winrt) | 15:01.43 |
mvrhel_laptop | Robin_Watts: I am fine with any name. | 15:02.00 |
Robin_Watts | and he'd prefer not to have the C++ guards in the headers. But I disagree with him. | 15:02.09 |
| I am happy to do the rewriting for you if you'd like. | 15:02.20 |
tor8 | the winRT sticks out too much and I've never liked mixed case filenames (too many woes of case sensitive vs insensitive file systems) | 15:02.25 |
mvrhel_laptop | Robin_Watts: feel free to rewrite. The issue there is that it was one file in with other files for that particular commit not sure how you revise that | 15:03.48 |
| Robin_Watts: I can't remember how we did the rename. Let me get my most recent stuff pushed up now | 15:04.07 |
Robin_Watts | mvrhel_laptop: That's what the 'e' entry in rebase -i is for :) | 15:04.17 |
mvrhel_laptop | and then if you want to rename it that is fine with me | 15:04.17 |
Robin_Watts | mvrhel_laptop: Go for it. | 15:04.33 |
mvrhel_laptop | ok hold on | 15:04.38 |
tor8 | Robin_Watts: the v in (v) is a bit thin IMO | 15:05.34 |
| paulgardiner: how about having the (+) on the first menu (not as a submenu) | 15:05.45 |
| also, maybe a fat check mark icon instead of "accept" to counterbalance the "x" on the left | 15:06.00 |
Robin_Watts | tor8: I can thicken it. | 15:06.15 |
| oh, I completely missed paulgardiner's new apk. Let me try it. | 15:07.10 |
tor8 | I think the main button order should be: link, reflow, outline, search, (+) | 15:07.22 |
| , (v) | 15:07.34 |
mvrhel_laptop | Robin_Watts: ok my recent changes have been pushed | 15:07.43 |
tor8 | that's a bit crowded on a phone ui though | 15:07.53 |
paulgardiner | tor8: Agree with order change, but the addition of (+) at top leve makes the lack of space problem worse | 15:08.39 |
mvrhel_laptop | the windows WebView UI object which one can use for html viewing is crap since I can't intercept any of the manipulations and it does not do reflow itself other than when a screen size change occurs. Zooming does no reflow. | 15:09.00 |
paulgardiner | tor8: Not sure what a fat check is, but yes an icon that suggest "Accept" in place of the word sounds good | 15:09.39 |
Robin_Watts | paulgardiner: A "tick" icon | 15:10.31 |
paulgardiner | Ah of course. | 15:11.01 |
| Yeah, like that. | 15:11.13 |
mvrhel_laptop | anyway I have to get kids to school. bbiaw | 15:11.27 |
Robin_Watts | mvrhel_laptop: will look in a mo. | 15:12.36 |
| paulgardiner, tor8: Fatter v in: http://ghostscript.com/~robin/toPaul.zip | 15:19.47 |
tor8 | there's a matching ic_check.png where we got the original icons | 15:21.58 |
paulgardiner | tor8: yeah, I'll grab it | 15:22.33 |
Robin_Watts | ah, that saves me drawing it :) | 15:22.51 |
| So, could we add some text to the top bar during such annotation things? | 15:23.45 |
| X freehand drawing TICK | 15:23.56 |
| So people know what mode they are in? | 15:24.14 |
tor8 | that's a good idea | 15:24.53 |
paulgardiner | You should get an info box come up, saying either "select text" or "draw annoation" | 15:25.03 |
tor8 | that one disappears though | 15:25.17 |
Robin_Watts | Yes, but it's nicer to have "highlight/underline/strike-out/ink" as appropriate at the top too. | 15:25.28 |
tor8 | now that each one has its own mode with X and Tick at the top, having the text at the top would be handy | 15:26.11 |
paulgardiner | I can try it I guess, but I think it might look odd. | 15:26.28 |
Sergey | Hello, does anyone know why (and what exactly does that mean) I can receive the Error: /invalidaccess in --run-- while trying to merge 2 pdf files: # gs -sDEVICE=pdfwrite -sOutputFile="/tmp/test.pdf" -f 1.pdf 2.pdf. Full error log here: http://pastebin.com/eXri1gGW | 15:30.02 |
kens | invalidaccess means you attempted to access an object which had permissions prohibiting it. | 15:30.44 |
| Try using a version of Ghostscript which is reasonably up to date, 8.70 is very old | 15:31.07 |
Sergey | Thanks, I'd try to find a newer in rhel repo. But just out of curiosity, is it possible to know what object that is gs fails to access? | 15:36.08 |
kens | possible, yes, easy , no. Its really not worth spending any time on this, the version is 4 years old and its 'probably' the use of CIDFonts. You really just need to use something new | 15:36.56 |
| Current version is 9.07 | 15:37.15 |
Sergey | I see, thanks again | 15:38.59 |
Robin_Watts | tor8: So, I've pushed the rewritten history of winRT to robin/winRT | 15:46.58 |
| I've renamed winRT to winrt, and I've removed all fiddling with the win32 solution. | 15:47.17 |
| I have not removed the C++ guards. | 15:47.28 |
| Other than the C++ guards, are you happy with it? | 15:54.00 |
| kens: if memory serves, in cases where the values of the Widths array differs from the values of the glyphs in the fonts, it's the Widths array that wins. | 16:04.50 |
kens | Yes | 16:04.58 |
| Basically that's the problem | 16:05.13 |
| The fact that its a CIDFOnt complicates matters further | 16:07.29 |
paulgardiner | Robin_Watts, tor8: http://intranet.glidos.net/~paul/MuPDF-new-menu-b.apk | 16:13.50 |
| Text between icons does seem to work nicely | 16:14.07 |
| I've also replaced the "Accept" by the check icon and reordered the top menu | 16:14.44 |
Robin_Watts | downloading now. | 16:16.02 |
| tor8: ping. | 16:27.18 |
| I'm being asked in #ubuntu-touch if we have a GTK or Qt based viewer for mupdf. | 16:27.49 |
| IIRC your new viewer used one of those, right? | 16:28.21 |
kens | Goodnight all | 16:36.49 |
Robin_Watts | paulgardiner: That looks luvverly. | 16:41.23 |
| The "Print failed" message could really do with being on screen longer. | 16:42.42 |
| but otherwise, that looks great. | 16:43.23 |
| I wonder if we should put titles in more menus. | 16:44.31 |
| Like "add annoation" in the (+) menu. | 16:44.48 |
chrisl | Huh, so reversing a block of bitmap data 64 bits at a time, valgrind throws "unitialised value" error. Traversing the same data, same address, 32 bits at a time, no error. | 17:17.34 |
Robin_Watts | chrisl: What version of valgrind ? | 17:22.50 |
| I have 3.8.1 built on peeves if you want me to try it with that. | 17:23.08 |
chrisl | I'm using 3.7.0 - have you found 3.8.1 produces fewer false positives? | 17:24.00 |
Robin_Watts | chrisl: I have not. | 17:24.17 |
| but it may be worth a try. | 17:24.25 |
chrisl | Okay, I'm using a small hack in gsalloc.c to force the block "large_size" to be 1 - like MEMENTO does | 17:25.38 |
Robin_Watts | or you can try for yourself on peeves; just put /home/robin/sauce/valgrind-3.8.1/bin/bin on the front of your path. | 17:25.43 |
chrisl | Er, okay I'll do that..... | 17:26.16 |
| Nope, 3.8.1 gives the same false error - oh well | 17:38.11 |
| Robin_Watts: thanks for the suggestion anyway | 17:38.25 |
Robin_Watts | so, bit reversal... how is that done? | 17:38.33 |
| valgrind has problems with some bitslicing operations. | 17:38.53 |
| If the 64 bit stuff actually compiles down to some 32bit ops with some bitshifts etc, then that might explain why valgrind has the problem. | 17:39.19 |
chrisl | The entire block of bitmap memory is memset() to 0x00 at allocation time, so there is no way in which the data is uninitialised | 17:39.56 |
Robin_Watts | unless something copies undefined bits in. | 17:40.23 |
| which is possible, but unlikely. | 17:40.34 |
chrisl | Then the read for the copy operation would/should show up | 17:40.53 |
Robin_Watts | not if the read doesn't change the flow of the program. | 17:41.21 |
chrisl | That still wouldn't explain why traversing the memory in "longs" throws the error, and traversing in "ints" doesn't | 17:42.15 |
Robin_Watts | chrisl: Presumably the code is doing *long_ptr = bit_reverse(*long_ptr) ? | 17:42.53 |
| it all depends how bit_reverse(long) is implemented. | 17:43.16 |
| but it seems like a valgrind bug, yes. | 17:43.39 |
chrisl | This isn't bit reversal - all it's doing it looking for a non-zero block of memory | 17:43.40 |
Robin_Watts | oh, you said "reversing" earlier. | 17:44.09 |
chrisl | oops, sorry - dropped the "t". I'm getting tired :-( | 17:44.31 |
Robin_Watts | Gotcha. | 17:44.37 |
| could it be that the long overflows the end of the buffer, where the int doesn't ? | 17:44.49 |
chrisl | No, we start at the end of the buffer and work back - and skip out before we get anywhere near the beginning of the buffer | 17:45.35 |
Robin_Watts | awesome. To build evince, I need gtk+-3.0 >= 3.7.5 | 17:46.07 |
| and the latest release version is... 3.6.4 | 17:46.22 |
chrisl | Yeh, I gave up trying to build evince, it was a pain! Why do you need to? | 17:46.48 |
Robin_Watts | chrisl: I was wondering how hard it would be to add mupdf as a backend. | 17:47.06 |
mvrhel_laptop | Hi Robin_Watts | 17:47.23 |
Robin_Watts | mvrhel_laptop: hi. I pushed an updated winrt branch onto robin/winRT | 17:47.44 |
chrisl | Robin_Watts: Oh, okay, you really need evince then. You could try an older release, I suppose | 17:47.58 |
Robin_Watts | chrisl: I'm git cloning gtk+ now. | 17:48.30 |
mvrhel_laptop | Robin_Watts: ok. so I guess I need to do a checkout of robin/winRT and then remove my winRT branch and then push winrt up to my repos? | 17:49.01 |
chrisl | Robin_Watts: building gtk+ from scratch (and getting it to work!) is likely to be a bit of a horror :-( | 17:49.13 |
Robin_Watts | mvrhel_laptop: safest way is to do: git branch myWinRT | 17:49.32 |
| (That will add a new branch "myWinRT" that will point to yours - in case we have an accident). | 17:50.00 |
| Then: git fetch robin winRT | 17:50.22 |
| (to pull my changes in as robin/winRT) | 17:50.34 |
mvrhel_laptop | ok trying this from git bash | 17:50.53 |
Robin_Watts | then while checked out on winRT, git reset --hard robin/winRT | 17:50.55 |
| All this assumes that you have no unsaved work. | 17:51.24 |
| uncommitted work, even. | 17:51.36 |
mvrhel_laptop | right. I had pushed everything up to the repos | 17:51.37 |
| Robin_Watts: is the command git reset --hard robin/winRT correct? | 17:58.19 |
Robin_Watts | yes. | 17:58.26 |
mvrhel_laptop | with the /? | 17:58.26 |
Robin_Watts | yes. | 17:58.31 |
| That says "reset the current branch (i.e. winRT) to be the same as robin/winRT" | 17:58.54 |
mvrhel_laptop | michaelv@XPS_LAPTOP /c/vrhel/Artifex_laptop/git_mupdf/mupdf (winRT) | 17:59.40 |
| $ git fetch robin winRT | 17:59.42 |
| remote: Counting objects: 83, done. | 17:59.44 |
| remote: Compressing objects: 100% (27/27), done. | 17:59.45 |
| remote: Total 77 (delta 54), reused 67 (delta 47) npacking objects: 37% (29/77) | 17:59.46 |
| Unpacking objects: 100% (77/77), done. | 17:59.48 |
| From ghostscript.com:/home/robin/repos/mupdf | 17:59.50 |
| * branch winRT -> FETCH_HEAD | 17:59.51 |
| michaelv@XPS_LAPTOP /c/vrhel/Artifex_laptop/git_mupdf/mupdf (winRT) | 17:59.53 |
| $ git reset --hard robin/winRT | 17:59.55 |
| fatal: ambiguous argument 'robin/winRT': unknown revision or path not in the wor | 17:59.56 |
| king tree. | 17:59.57 |
| Use '--' to separate paths from revisions, like this: | 17:59.59 |
| 'git <command> [<revision>...] -- [<file>...]' | 18:00.00 |
| michaelv@XPS_LAPTOP /c/vrhel/Artifex_laptop/git_mupdf/mupdf (winRT) | 18:00.02 |
| $ | 18:00.04 |
Robin_Watts | OK, then git reset --hard e989a0e | 18:00.23 |
mvrhel_laptop | ok that has the Rename winRT to winrt | 18:01.02 |
Robin_Watts | perfect. | 18:01.12 |
mvrhel_laptop | so now do I push up to my repos? | 18:01.25 |
Robin_Watts | At no point in the history have any of those commits changed anything to do with win32. | 18:01.34 |
mvrhel_laptop | oh great | 18:01.46 |
Robin_Watts | sure. git push -f origin winRT | 18:01.46 |
mvrhel_laptop | so is the branch winRT now or winrt? | 18:02.40 |
Robin_Watts | The branch is still winRT | 18:06.22 |
| The directory should be winrt though. | 18:06.36 |
mvrhel_laptop | so a few comments about the questions that Ovitz had. I wonder if he was hoping to use the library with calls from JavaScript or .NET applications. That can be done but requires a little wrapping of the methods so that they are true windows runtime components. Right now, this is all written in c++/xaml which is the recommended way for windows 8 apps. But it is possible to set up... | 18:07.08 |
| ...things so that users can make rendering calls from .NET. I really don't feel like drinking that much microsoft kool-aid | 18:07.10 |
| It he comes back, and I am not here, find out what type of language he is working with | 18:07.57 |
| s/It/If/ | 18:08.02 |
| I suppose the way I should do this though is to wrap everything up so that they can be used. I just don't want to write a .NET viewer | 18:08.49 |
Robin_Watts | mvrhel_laptop: I understand. | 18:17.27 |
| I would love to wrap the MuPDF calls so that they can be called from java directly, so they'd be accessible to non-C programmers on android. | 18:18.14 |
| but that's a lot of work. | 18:18.20 |
| I think .net would be the same. | 18:18.28 |
| mvrhel_laptop: So, his compile failed... is there a file missing ? | 18:19.13 |
| I don't have Windows 8 here to try. | 18:19.31 |
mvrhel_laptop | Robin_Watts: yes. until the last push I did the compile would have failed | 18:19.50 |
| when I did the sync up with the golden I was missing 2 files | 18:20.18 |
Robin_Watts | right. | 18:20.24 |
mvrhel_laptop | plus there was an issue with fopen | 18:20.31 |
| that is all fixed now | 18:20.39 |
Robin_Watts | oh, you mean there were files missing from the solution, so symbols were missing ? | 18:20.56 |
mvrhel_laptop | yes | 18:21.02 |
Robin_Watts | He was complaining of an actual missing file. | 18:21.05 |
mvrhel_laptop | oh | 18:21.08 |
| let me do a clean check out | 18:21.14 |
| and make sure that is not happening | 18:21.21 |
| brb | 18:21.32 |
Robin_Watts | LVContents.h/cpp | 18:21.51 |
mvrhel_laptop | oh crap | 18:22.03 |
| yes. that needs to be added | 18:22.08 |
| ok | 18:22.13 |
| sigh | 18:22.22 |
Robin_Watts | tor8: ping | 18:23.35 |
| At the moment, we have fz_images and pdf_images. | 18:23.50 |
| In order to be able to write out images without decompressing/recompressing them, I need to pull more of the pdf_image stuff into the fz_image stuff. | 18:24.26 |
| It looks like there is no particular pdf specific stuff in pdf_images, so I may just move pdf_images to be fz_images. | 18:25.33 |
tor8 | Robin_Watts: sure. | 19:17.32 |
Robin_Watts | tor8: Did you see my earlier burblings about GTK? | 19:18.00 |
| #ubuntu-touch-meeting in 10 minutes for those interested. | 19:19.18 |
| tor8: Was your viewer GTK+ based? | 19:29.05 |
tor8 | Robin_Watts: yes. | 19:33.46 |
sebras | Robin_Watts: they seem to be awfully quiet over at ubuntu-touch though..? | 20:22.36 |
sebras | is not a lurker _at all_... ;) | 20:23.00 |
Robin_Watts | ubuntu-touch-meeting ? | 20:24.17 |
sebras | Robin_Watts: oh. ?-) | 20:24.42 |
Robin_Watts | Not a huge amount was discussed. There never seems to be a quorum of developers at the meeting. | 20:25.12 |
sebras | oh. | 20:26.08 |
| I got notified about discussions about a ubuntu-touch client and my name, hence my interest. | 20:28.02 |
mvrhel_laptop | Robin_Watts: you still up? | 23:23.19 |
Robin_Watts | I am | 23:23.25 |
mvrhel_laptop | so even though my branch shows it is up to date with the winrt commit that you did, my folder is still winRT | 23:23.57 |
Robin_Watts | mvrhel_laptop: OK. | 23:24.19 |
mvrhel_laptop | should it not have been renamed? | 23:24.32 |
Robin_Watts | Any changes in your working set? | 23:24.39 |
mvrhel_laptop | I was about to get started on some stuff | 23:24.51 |
Robin_Watts | If not, rm -rf winRT then git checkout winrt | 23:24.57 |
mvrhel_laptop | but I can hold off for a bit to get this set | 23:25.00 |
Robin_Watts | I suspect it's just because windows is case insensitive | 23:25.11 |
mvrhel_laptop | ok let me do that | 23:25.11 |
| hmm with the check out it did D of all the files in winrt | 23:28.32 |
| and now there is not a winrt nor a winRT | 23:28.59 |
Robin_Watts | mvrhel_laptop: Hmm. Let me try and fix it. | 23:29.15 |
| how odd. It seems to be fine for me. | 23:29.55 |
| try: git reset --hard ? | 23:30.03 |
mvrhel_laptop | ok hold on | 23:30.08 |
| that got me back to winRT | 23:30.36 |
| the branch name is winrt though | 23:31.16 |
Robin_Watts | ? | 23:31.23 |
mvrhel_laptop | it was winRT | 23:31.35 |
| now it is winrt | 23:31.45 |
Robin_Watts | git checkout master | 23:32.08 |
| and check that the winRT/rt directory has gone. | 23:32.20 |
| then git checkout winrt | 23:32.36 |
| Oh! | 23:32.52 |
| I think I see what's going on. | 23:33.30 |
mvrhel_laptop | ok in master no winrt/RT directory | 23:33.31 |
Robin_Watts | now: git checkout winRT | 23:33.43 |
mvrhel_laptop | ok now I have a winRT directory | 23:34.01 |
Robin_Watts | What hash are you on ? | 23:34.34 |
mvrhel_laptop | 5043f2a54e6e13bedcea04727d440022931ffed3 | 23:35.08 |
| from when I added the missing files | 23:35.22 |
Robin_Watts | Right. I was expecting you to be on e989a0e | 23:35.34 |
mvrhel_laptop | I had not noticed the folder when I did the missing file commit. Is that the issue? | 23:36.28 |
| did I somehow set it back :( | 23:36.59 |
Robin_Watts | Oh, urm, is 5043f2 the one after e989a0e ? | 23:36.59 |
mvrhel_laptop | yes | 23:37.08 |
| and the is where my local and the repos on casper is | 23:37.21 |
Robin_Watts | just a tick. | 23:37.23 |
mvrhel_laptop | is shows I added the files to the winRT folder in 5043f2 | 23:37.51 |
Robin_Watts | yes, that'll be it. | 23:38.33 |
mvrhel_laptop | so obviously my directory structure was wacked at that time | 23:38.33 |
Robin_Watts | Just a mo. | 23:38.35 |
| ok, so: git reset --hard HEAD~1 | 23:40.32 |
| to throw away that last commit | 23:40.38 |
| then: git pull --rebase robin winRT | 23:40.48 |
| With luck you should be on 9baa0cf | 23:41.08 |
| with a winrt directory | 23:41.18 |
| if not you'll need to checkout to master and back. | 23:41.32 |
| I should probably have rewritten the history so that it was always called winrt. | 23:44.08 |
mvrhel_laptop | ok let me try | 23:44.15 |
| should I be in winRT branch? | 23:44.29 |
Robin_Watts | Yes. | 23:44.34 |
| Then git reset --hard HEAD~1 will throw away the top commit | 23:44.51 |
| and git pull --rebase robin winRT will pull in my replacement commit. | 23:45.09 |
mvrhel_laptop | ok I am on 9baa0cf | 23:45.43 |
| but with winRT as my directory | 23:45.49 |
Robin_Watts | Then that may leave you with a winRT directory temporarily. To cure that you git checkout master then git checkout winrt | 23:45.50 |
| git checkout winRT, sorry. | 23:45.59 |
tor8 | Robin_Watts: git filter-branch can rename directories | 23:46.14 |
mvrhel_laptop | ok. hold on. so I did a git reset --hard HEAD~1 | 23:46.32 |
| and then git pull --rebase robin winRT | 23:46.40 |
Robin_Watts | right. | 23:46.47 |
| now you and I are both on 9baa0cf | 23:46.57 |
| but you still have the old winRT directory from before. | 23:47.15 |
mvrhel_laptop | yes | 23:47.18 |
Robin_Watts | so git checkout master will get rid of it. | 23:47.24 |
| and then git checkout winRT will recreate it with the right name. | 23:47.39 |
mvrhel_laptop | yes! | 23:48.40 |
Robin_Watts | excellent. | 23:48.44 |
mvrhel_laptop | thanks Robin_Watts | 23:48.47 |
Robin_Watts | no worries. | 23:48.51 |
mvrhel_laptop | sorry to bother you with all of this | 23:48.55 |
Robin_Watts | git is fun! | 23:49.04 |
mvrhel_laptop | like going to the dentist | 23:49.11 |
| actually it is very powerful | 23:49.26 |
| and I am starting to feel more comfortable with it | 23:49.44 |
Robin_Watts | The ability it has to rewrite history is hugely powerful. | 23:49.47 |
mvrhel_laptop | have you ever heard of an eglu? It is a British product | 23:50.11 |
Robin_Watts | mvrhel_laptop: The trick is to get away from thinking of it as "a better cvs". | 23:50.15 |
| Don't eskimos live in them? :) | 23:50.23 |
mvrhel_laptop | well not these. | 23:50.35 |
Robin_Watts | Having googled, I can report that a friend keeps rabbits in one. | 23:50.56 |
mvrhel_laptop | wife is wanting chickens. and I am either going to build a coop or buy a refab one | 23:50.57 |
| oh ok | 23:51.04 |
| s/refab/prefab/ | 23:51.17 |
Robin_Watts | ah, yeah. We have foxes here. | 23:51.25 |
mvrhel_laptop | we had a coyote in our backyard yesterday | 23:51.38 |
| so I need something secure | 23:51.42 |
| he was good sized too | 23:51.53 |
| and surprisingly out in the middle of the day | 23:52.11 |
| but looking healthy which explains why I dont see any rabbits around here anymore | 23:52.30 |
Robin_Watts | Hmm. Mr Google says eglu's are $500 or so here. | 23:52.32 |
mvrhel_laptop | oh wow that is way cheaper | 23:52.40 |
tk111 | hi, could someone please give an advice how to convert tiff to pdf using gs? | 23:52.49 |
Robin_Watts | http://www.harrodhorticultural.com/eglu-go-chicken-house-and-2m-run-pid9004.html?Aff=GPL&gclid=CKWUjdSasrYCFUbKtAodmhsAWg | 23:52.49 |
| mvrhel_laptop: What size are you looking at? | 23:53.12 |
mvrhel_laptop | http://www.omlet.us/products_services/products_services.php?cat=Eglu+Cube&subcat=The+Eglu+Cube | 23:54.42 |
| that might be a bit big actually | 23:55.16 |
| I am only going to have 3 to 4 chickens | 23:55.29 |
Robin_Watts | I fear that'll be a bit big for me to carry :) | 23:55.57 |
mvrhel_laptop | hehe | 23:56.01 |
| actually this might work just fine http://www.omlet.us/products_services/products_services.php?cat=Eglu+Go&subcat=The+Eglu+Go. need to talk with the wife | 23:57.25 |
Robin_Watts | £699 over here. | 23:57.58 |
mvrhel_laptop | and you can buy it with 2 chickens. ok it is cheaper here | 23:58.07 |
Robin_Watts | Right. The go is £350 quid here - so $500 or so? | 23:58.43 |
mvrhel_laptop | yes $495 | 23:58.55 |
| ok. time to beat on windows a bit | 23:59.46 |
| Forward 1 day (to 2013/04/05)>>> | |