| <<<Back 1 day (to 2015/07/30) | 20150731 |
sebras | Robin: does a patch like the one over at sebras/signedness fix any of the issues for you? while I see warnings in load-gif.c when using -Wsign-conversion and -Wsign-compare I _also_ see similar problems in load-png.c so I'm a bit confused as to what warnings _you_ see. do you filter out the ones in load-png.c? | 00:10.05 |
Robin | 4>c:\artifex\mupdf.git\source\fitz\load-gif.c(166) : warning C4018: '<' : signed/unsigned mismatch | 00:57.09 |
| 4>c:\artifex\mupdf.git\source\fitz\load-gif.c(197) : warning C4018: '<' : signed/unsigned mismatch | 00:57.10 |
| 4>c:\artifex\mupdf.git\source\fitz\load-gif.c(214) : warning C4018: '<' : signed/unsigned mismatch | 00:57.12 |
| 4>c:\artifex\mupdf.git\source\fitz\load-gif.c(531) : warning C4018: '<' : signed/unsigned mismatch | 00:57.13 |
| 4>c:\artifex\mupdf.git\source\fitz\load-gif.c(532) : warning C4018: '<' : signed/unsigned mismatch | 00:57.15 |
| just those. | 00:57.19 |
| I don't filter any warnings here. | 00:57.48 |
sebras | Robin: I have a new proposed patch over at sebras/signedness, do you mind testing it? | 09:04.46 |
| actually anyone that has the possibility to build mupdf in visual studio could test it, so robin doesn't have to work on every day of his vacation. ;) | 09:05.41 |
chrisl | sebras: Robin doesn't do vacations as the rest of us understand them..... and I barely understand the concept myself. | 09:06.40 |
| The good news is that the github mupdf mirror is working :-) | 09:08.20 |
sebras | chrisl: well, when I'm on vacation I submit patches to mupdf, when I work I submit to gstreamer. my friends are worried for my sanity. | 09:10.09 |
chrisl | sebras: we'd be much more worried if you were submitting to gstreamer when on vacation! | 09:11.20 |
| sebras: I see one signed/unsigned warning remaining - on line 197 | 09:23.44 |
sebras | chrisl: ok, updated patch. | 09:25.36 |
| chrisl: not seeing the warnings myself make it a bit more problematic to test. :) | 09:26.02 |
chrisl | sebras: it's building now (takes a bit 'cause it's in a VM across a net share). If this doesn't solve it, I'll poke at it, and throw something up for you to see | 09:27.45 |
sebras | chrisl: cool, thanks. :) | 09:28.33 |
chrisl | sebras: yep, that gif code builds cleanly in VS2005 | 09:32.18 |
| I make no comments or claims about the correctness - I'll leave that for tor8! | 09:32.43 |
tor8 | chrisl: the patch looks reasonable | 09:33.31 |
| chrisl: thanks for testing. I never got around to reinstalling MSVC on my windows machine... | 09:34.06 |
chrisl | I didn't even look at it, I just through it at VS..... | 09:34.06 |
chrisl | suddenly realises he's voluntarily using Visual Studio.... realises something must be wrong, and goes in search of coffee....... | 09:34.31 |
sebras | chrisl: thanks! :) | 09:37.21 |
| tor8: patch rebased on origin/master and period added at end of commit message... jeez! will I ever learn?! | 09:40.48 |
chrisl | sebras: no problem | 09:42.06 |
tor8 | sebras: you could add a commit hook to your local repository to check for final punctuation ;) | 09:48.04 |
Mo | Hi. I like mupdf for years now, also on Android. Android has the nice feature to skip forward by tapping right side. Does the desktop Mupdf also have something like this? | 09:56.54 |
| <Mo> Because if not full screen, page down skips pages but doesn't scroll. Android does both scrolling automatically and jumping to the next page. | 09:56.54 |
tor8 | Mo: the up/down arrow keys will flip pages when you reach the top/bottom of the page | 10:03.38 |
sebras | tor8: hm.. that actually sounds like a good idea. :) | 10:21.10 |
Mo | tor8: up/down doesn't do the same as it does not scroll page wise but line-wise or something. Then I'm not sure then the page break takes place when scrolling. | 10:40.46 |
| Did you try the Android behaviour? It is special, useful, have never seen that on other viewers. | 10:41.11 |
tor8 | Mo: yes. I've replicated the android behaviour in a new prototype desktop viewer that uses OpenGL | 10:46.58 |
| that's not merged into the main branch, but if you want to try it you can find it on my personal repo in the 'glut' branch | 10:47.20 |
| http://git.ghostscript.com/?p=user/tor/mupdf.git;a=shortlog;h=refs/heads/glut | 10:47.41 |
Mo | tor8: Thanks. Did you change that right now or was it your own idea to copy that behaviour? | 10:57.17 |
| tor8: Which change has done this? | 10:58.58 |
tor8 | Mo: http://git.ghostscript.com/?p=user/tor/mupdf.git;a=commitdiff;h=cd18b0fc3e9912c25507fb874a33699296987c9e;hp=30ffc99d5939aaa8c7e39ffb8e1c97af4d0b76c4 | 10:59.39 |
Mo | tor8: Interesting, this is identical to the Android code? | 11:07.52 |
| It's committed some while ago, Will that ever go to the release? | 11:08.08 |
| tor8: I cloned git://git.ghostscript.com/user/tor/mupdf.git, there is only master branch. Is that the right branch? | 11:18.35 |
| I have built mupdf/build/debug/mupdf-x11 but space only behaves like PageDown. | 11:22.09 |
tor8 | Mo: you need to build mupdf-glut from that branch | 12:41.09 |
Mo | tor8: How? | 12:59.52 |
tor8 | if you have checked out the 'glut' branch, it should get built by default | 13:00.54 |
Mo | tor8: git branch --list only says * master | 13:01.48 |
tor8 | git remote add tor git://git.ghostscript.com/user/tor/mupdf.git | 13:02.06 |
| git remote update | 13:02.09 |
| git checkout tor/glut | 13:02.12 |
kens | henrys_ when you wake up and can you ping me please | 13:39.08 |
chrisl | kens: I think henrys_ is off today | 13:40.04 |
kens | Oh drat | 13:40.09 |
| I implemented the 'fix' for regular text so that it behaves the same as with -dTextAlphaBits=4 | 13:40.35 |
| WHicih turned up 3 failing PDF file in our test suite, whcih I fixed. | 13:40.50 |
| However, there are two PCL files which show the same problem. | 13:40.58 |
| THat is, if you run them with -dTextAlphaBits, they throw an error | 13:41.14 |
chrisl | No current point again? | 13:41.35 |
kens | THe matrix is not invertible, so currentpoint filas | 13:41.50 |
chrisl | So, could we use a different method to get the current point? | 13:42.58 |
kens | I'm not certain about that.. And in any event, the PostScript case *should* fail, according to testing | 13:43.30 |
| I cna commit the PDF interpreter fixes anyway, I was just wonderign whether I should open a bug report for the PCL files and wait for Henry to fix them, or just commit the change to cause the error anyway. | 13:45.31 |
| TO be honest, I don;t thnk the PCL interpreter should be setting the CTM to 0 0 0 0 Tx Ty | 13:47.57 |
chrisl | That does seem bad.... | 13:48.35 |
kens | It seems the PCL scale is 0 0 | 13:48.47 |
| So just not drawing the text would seem to be favourite to me | 13:49.00 |
| Assuming that PCL interpreters don't throw an error on that, whch seems like they don't because ths is a QL test file | 13:49.36 |
| and it in the middle of the teest page | 13:49.46 |
| somethign called pfp->params.height_4ths is 0, that's the root cause. | 13:51.21 |
henrys_ | kens I can just fix it after you commit. I thought I'd gotten rid of all the singular matrix issues. ;-( | 13:51.38 |
kens | henrys_ if you're happy with that I'll go ahead. | 13:51.53 |
henrys_ | yup | 13:52.05 |
kens | OK Shall I open you a bug reprot ? | 13:52.12 |
| report* | 13:52.16 |
henrys_ | yea I'm off today and Monday so forgetting is likely otherwise | 13:52.38 |
kens | Right, consider it done and have a grea\t weekend! | 13:52.49 |
henrys_ | thanks | 13:53.04 |
chrisl | henrys_: when/where is the race? | 13:53.26 |
henrys_ | I'm at home for this one: http://www.ironman.com/triathlon/events/americas/ironman/boulder.aspx#axzz3hTdshLjU | 13:54.51 |
kens | So no sea level advantage then, though I guess you're still up on any sea level athletes anyway | 13:55.29 |
chrisl | Hope it goes well! | 13:55.40 |
kens | 112 miles..... | 13:56.09 |
| I feel weak just reading about it | 13:56.18 |
henrys_ | elevation and our dry heat definitely a problem but I've got a leg up on the out of towners | 13:56.59 |
| swimmer in particular have problems with the elevation | 13:57.27 |
rayjj | I wondered why the profiler run was so fast, but it was because it reset back to the default "sampling" instead of instrumentation. VS instrumentation profiling is *way* slower than running a profle build on linux. On linux, the impact is about 15%, with VS2015 it's almost 400% | 16:37.07 |
mvrhel_laptop | bbiaw | 17:35.24 |
| Forward 1 day (to 2015/08/01)>>> | |