| <<<Back 1 day (to 2016/11/13) | 20161114 |
tor8 | Robin_Watts: so, a few commits on tor/master | 12:05.53 |
| one to add back link target page coordinates; this time properly transformed (and working for EPUB as well) | 12:06.15 |
malc_ | woohoo! | 12:07.46 |
Robin_Watts | tor8: Nice. Let me look. | 12:08.12 |
| It would be really nice to add some API comments for stuff like pdf_add_stream etc. | 12:09.48 |
| Other than that they look fine to me. | 12:11.50 |
| 3 commits on robin/master. I am testing the 3rd one now. | 12:12.30 |
sebras | tor8: links in epub has no x-coordinate? | 12:14.32 |
tor8 | sebras: it does not | 12:14.43 |
| (it defaults to 0 in the fz_resolve_link function, if you're wondering why I don't set it in epub_resolve_link) | 12:15.14 |
sebras | tor8: yeah, I gathered that after taking a quick glance, but I wanted to confirm that this was correct. :) | 12:15.53 |
| I'm holding on to my commits until 1.10 is out. | 12:16.17 |
tor8 | there were a couple of bug reports on 1.10-rc1 | 12:16.37 |
sebras | tor8: oh.. so the epub commit is a fix for 697338? | 12:18.32 |
tor8 | sebras: s/epub/link target/ | 12:19.12 |
sebras | right. | 12:19.23 |
| tor8: do we want to track that in the commit message? | 12:19.34 |
tor8 | no. there are more than the one bug. | 12:20.17 |
sebras | ah. | 12:20.25 |
tor8 | and I wrote this commit *before* the bugs came in :) | 12:20.39 |
sebras | aha. | 12:23.08 |
| tor8: I tested to update to curl 7.32.0. there is an immediate segfault. | 12:51.02 |
tor8 | we don't use curl in the default build, so I'm not too stressed about the curl vulnerabilities | 12:56.52 |
| but if you can get them fixed, that's still good | 12:57.01 |
sebras | tor8: I know. | 13:14.30 |
| tor8: ugh... something appears to be awry with the threading. | 13:33.17 |
| tor8: I found the curl commit that provokes the change, but I cannot see deprecating a progress callback that we aren't using should provoke the stream session handle to suddenly become invalid. | 13:34.07 |
tor8 | oh crap. the form/annotation event handling conflicts with simple link handling... | 13:52.00 |
Robin_Watts | Ok, my layertest tool shows everything is working OK. | 14:08.04 |
| So 6 commits on robin/master. Ignore the last one, and probably the last but one too. | 14:08.36 |
| The others probably aren't for 1.10, but are for master. | 14:08.56 |
| (although, arguably the "Workaround 0 line-width" one could go into 1.10) | 14:09.23 |
| presumably 1.10 is being done from a branch? | 14:09.33 |
Robin_Watts | lunches | 14:09.36 |
sebras | tor8: ok, so I have succeeded in updating curl localy I think. | 14:44.53 |
| tor8: something messed up with cached object files so I opted to make clean between each build. | 14:45.32 |
| tor8: doing so resolved all my spurious issues. | 14:45.41 |
| tor8: how do I go about pushing to the thirdparty curl repo? | 14:45.55 |
Robin_Watts | sebras: It's just a repo on casper. | 14:46.25 |
sebras | Robin_Watts: ok. | 14:46.37 |
| Robin_Watts: what about the origin/mupdf branch? | 14:46.44 |
| do I merge curl-7_51_0 to that one? | 14:46.53 |
| I have been using rebase locally, but I suppose that this doesn't fly when doing this officially. | 14:47.15 |
Robin_Watts | sebras: I'm confused. | 14:47.28 |
| The curl repo is a clone of the official curl repo, right? | 14:47.47 |
sebras | Robin_Watts: yes, but we have local patches. | 14:48.19 |
tor8 | yes. doing a 'git remote update' in /home/git/thirdparty/curl.git should do it | 14:48.28 |
sebras | Robin: so when upstream updates, what do we do..? | 14:48.30 |
Robin_Watts | so, yes, I see, push the new curl repo, and then merge to the mupdf branch. | 14:48.32 |
| sorry, was slow on the uptake there. | 14:48.44 |
tor8 | sebras: I have updated casper's thirdparty/curl.git | 14:48.55 |
Robin_Watts | tor8: I need to get these layer changes to Raed, so I need to commit them, I think. | 14:50.38 |
tor8 | sebras: to update the mupdf branch you should merge (so we don't garbage collect old sha1's) | 14:50.51 |
Robin_Watts | Presumably we are doing a 1.10 branch rooted at 1.10rc and then cherry picking any patches we need into that? So development can continue on master as normal ? | 14:51.12 |
tor8 | and push to the curl.git | 14:51.13 |
sebras | tor8: yeah, that made the most sense, but no one had updated before so I had no pattern to follow. | 14:51.16 |
tor8 | Robin_Watts: either that, or taking what we got when we've fixed al lthe 1.10rc1 issues and make an rc2 | 14:51.43 |
| sebras: I make a separate repo clone when working on thirdparty modules, so I don't accidentally mess up and commit a submodule checksum that hasn't been pushed upstream yet | 14:52.39 |
| or you can work in the submodule checkout by fudging with the remote set-url | 14:53.02 |
| but that is fraught with dangers | 14:53.10 |
sebras | this is of course exactly what I did. | 14:53.29 |
| to my surprise, I apparently like danger. | 14:53.46 |
| tor8: ok, two patches on sebras/master | 15:02.01 |
| tor8: I clustered it successfully in my area. | 15:05.08 |
tor8 | sebras: did you push the curl git to mupdf.git? | 15:07.28 |
| the 'mupdf' branch on mupdf.git has the whole curl history | 15:07.50 |
sebras | awww.. crap. | 15:07.56 |
| yes. :-( | 15:08.02 |
Robin_Watts | git push origin :mupdf | 15:08.52 |
tor8 | I already zapped it | 15:09.02 |
Robin_Watts | ok. | 15:09.06 |
sebras | tor8: thanks. | 15:09.07 |
tor8 | but we probably ought to 'git gc' on casper so people don't need to download all of curl's history when they clone the repo | 15:09.35 |
sebras | tor8: thanks for fixing my mistake, hopefully I have now pushed to the correct repo. | 15:11.59 |
| tor8: I don't trust me with running 'git gc' on casper right now. :) | 15:12.22 |
Robin_Watts | Updated commits on robin/master | 15:12.45 |
sebras_ | tor8: sorry, I dropped out when my laptop died without warning. did the patches look reasonable? | 15:19.30 |
kens | Hmm sebras' laptop looks like its in trouble.... | 15:23.31 |
sebras | kens: naa, I just disabled my battery indicator and accidentally unplugged the charger... | 15:23.56 |
kens | It was sebras_ disconnecting that made me thik you had a problem | 15:24.29 |
tor8 | sebras: I've tweaked the ghostscript.com/thirdparty/curl.git repo to properly update like a mirror if you do 'git remote update' in it | 15:24.38 |
| it was missing some configuration entries to behave properly like a mirror repo | 15:25.11 |
sebras | tor8: cool, that makes it easier to keep updated. | 15:25.34 |
tor8 | I don't see anything objectionable, but the curl.vcproj needs updating | 15:26.42 |
sebras | tor8: do we compile curl in android too? | 15:27.08 |
tor8 | I don't know | 15:27.25 |
| certainly not for the new JNI stuff | 15:27.39 |
| the old viewer, I don't know | 15:27.45 |
Robin_Watts | We do not. | 15:27.49 |
| mupdf-curl is the only thing curl is used for - so windows/linux is all. | 15:28.12 |
tor8 | Robin_Watts: glyph-cache.h still refers to fz_pixmap and fz_matrix so shouldn't it still include math.h and pixmap.h? | 15:30.24 |
Robin_Watts | Yes. | 15:31.19 |
| will fix. | 15:31.22 |
tor8 | draw-glyph.c includes fitz.h, draw-imp.h and glyph-cache-imp.h ... should it not include fitz-imp.h instead of mupdf/fitz.h like the other files? | 15:33.08 |
| how do we decide when to put things in fitz-imp.h and other-imp.h? | 15:33.46 |
Robin_Watts | stuff like draw-imp.h is stuff that's just needed for the drawing routines. | 15:34.12 |
| fitz-imp.h is for "anything that doesn't have it's own blah-imp.h" | 15:34.47 |
tor8 | "yet" :) | 15:35.01 |
Robin_Watts | indeed. | 15:35.07 |
| We should do separate compilation tests on every header. | 15:35.31 |
| tor8: draw-glyph.c doesn't include fitz-imp.h cos it doesn't need anything that's in fitz-imp.h | 15:36.44 |
tor8 | in pdf_set_layer_config_as_default, when copying the order and rbgroups into the configs | 15:40.31 |
| the comment says "copy when the config doesn't have one already" | 15:40.42 |
| but the test seems to be the opposite | 15:40.47 |
| if (order && pdf_dict_get(Order)) pdf_dict_put(Order, order) | 15:41.04 |
Robin_Watts | tor8: oops. | 15:43.56 |
| Fixed version online. | 15:46.41 |
| Commit 2216f940 introduces a problem in pdf-page.c at line 512. | 15:57.19 |
| return pdf_page_obj_transform(...); is bad, cos it's a void. | 15:57.38 |
tor8 | Robin_Watts: d'oh! | 16:09.34 |
| Robin_Watts: fix for that (and a couple of other things) on tor/master | 16:10.50 |
| Robin_Watts: your commits up to and including 'Continued Layer API work.' LGTM | 16:11.35 |
| simple layertest file is missing layertest.vcproj | 16:11.56 |
| and is resorting the project parts for some reason | 16:12.17 |
Robin_Watts | Thanks. | 16:13.29 |
| I'm not sure I understand the changes to fz_is_external_link. | 16:14.45 |
| Otherwise all look good (apart from the fact you have a WIP commit in the middle) | 16:15.25 |
tor8 | Robin_Watts: mailto:foo@bar.com is also an external link, not just http:// and file:// etc | 16:18.28 |
Robin_Watts | Ah, ok. | 16:18.56 |
| tor8, sebras: There is an MSVC change on robin/master that fixes a few problems with building the commercial builds under MSVC. | 18:18.08 |
| No code changes, just solution tweaks. | 18:18.24 |
sebras | Robin_Watts: how does platform/win32/mupdf-curl.vcproj find all the curl files? | 18:24.12 |
| Robin_Watts: I can see that it picks up the ones in platform/x11 and curl_stream.* but what about the files in thirdparty/curl/..? | 18:24.47 |
Robin_Watts | sebras: It depends on libcurl.vcproj :) | 18:24.55 |
sebras | Robin_Watts: how can I know that? | 18:25.11 |
Robin_Watts | which is in thirdparty/curl/vs/vc80/ or something. | 18:25.15 |
sebras | Robin_Watts: oh. | 18:25.25 |
Robin_Watts | That's in mupdf.sln in the unreadable dependencies blocks :) | 18:25.40 |
| thirdparty\curl\vs\vc8\lib\vc8libcurl.vcproj | 18:26.14 |
sebras | Robin_Watts: do you mind testing sebras/master in MSVC? | 18:26.54 |
Robin_Watts | sebras: Sure, but in a bit. | 18:27.05 |
sebras | I want to make sure I don't mess anything up with the curl 7.51.0 update. | 18:27.11 |
| Robin_Watts: ok, I might not be around to fix it later, but in that case I' | 18:27.27 |
Robin_Watts | I have been trying to debug something for ages, but the luratech builds kept failing and screwing it up. | 18:27.33 |
sebras | ll take a look at it tomorrow or the day after. | 18:27.36 |
| Robin_Watts: np. | 18:27.51 |
sebras | sleeps | 18:47.36 |
| Forward 1 day (to 2016/11/15)>>> | |