| <<<Back 1 day (to 2014/02/18) | 2014/02/19 |
Robin_Watts | henrys: yup. | 00:13.13 |
ray_laptop | Hi all. Can someone review and approve my change to fix 3 bugs (one customer bug) ? This adds a pngalpha_put_image function and looks OK. | 00:44.59 |
Robin_Watts | peers at it. | 00:45.24 |
ray_laptop | Thanks to mvrhel_laptop (I think, according to blame) for the bit_put_image I based it on. | 00:45.54 |
Robin_Watts | borrwed | 00:46.41 |
ray_laptop | I looked at the results from mvrhel_laptop's Transparency-DesignGuide.pdf and it looks OK. | 00:47.09 |
| Robin_Watts: oh, a typo in the commit log > | 00:47.20 |
Robin_Watts | In a comment. | 00:48.11 |
| I'm intrigued by alpha_plane_index | 00:48.25 |
| You insist on num_chan == 3 | 00:48.37 |
| and you insist that alpha_plane_index > 0 | 00:48.49 |
ray_laptop | Robin_Watts: found the typo. Thanks | 00:48.55 |
Robin_Watts | but then the look basically assumes it's 3. | 00:49.01 |
| s/look/loop/ | 00:49.07 |
| as 0 1 and 2 are used for r/g/b | 00:49.22 |
ray_laptop | Robin_Watts: the loop can then assume its 3 since its been checked for | 00:49.41 |
Robin_Watts | oh, might there be a shape plane too? | 00:49.42 |
| Surely the if should be: alpha_plane_index < 3 ? | 00:50.13 |
ray_laptop | but the component ordering corresponds to the pngapha encode_coloe | 00:50.14 |
| color | 00:50.16 |
| The checks are rather superfluous | 00:50.46 |
Robin_Watts | but it looks sensible enough to me. | 00:51.00 |
ray_laptop | As I waid -- I borrowed from the bit_put_image code | 00:51.34 |
| Robin_Watts: but I agree, and will change, to look for alpha_plane_index < 3 since we do assume that | 00:52.32 |
| Robin_Watts: thanks for the PROMPT review | 00:53.00 |
Robin_Watts | np. | 00:53.10 |
| I'm struggling to understand the values coming out of the picsel font engine... this is hurting my head | 00:53.35 |
ray_laptop | I pity you. Fonts are NOT my favorite thing to deal with. clist is bad enough | 00:55.59 |
Robin_Watts | The annoying thing is that this is not even outlines. This is just metrics. | 00:57.24 |
ray_laptop | Robin_Watts: metrics, IMO, are more challenging than outlines. You can be "close enough" on an outline, but if the metrics are wrong, the entire layout suffers. It's hare to spot an outline difference (in general) | 02:56.59 |
| where a glyph is placed in the font bbox is even less critical since you can always get that by using a large bbox and detecting the "marked" exents | 02:58.16 |
| but the glyph spacing/kerning and line-to-line spacing seem mysterious to me. | 02:59.13 |
| they never seem to match my hand calculated values even for somewhat simple Type 1 fonts that I've delved inot | 02:59.54 |
| s/inot/into/ | 03:00.01 |
NP-Hardass | I'm a postscript newbie. I just attempted "200 200 moveto currentpoint pstack" and the stack shows 200.0 200.000031 Non integers in this case confuses me. Any ideas why this is the case? | 03:18.29 |
Raymond_ | Hello? | 10:37.09 |
| when to use "FZ_IMAGE_PNG" ? | 10:37.58 |
Robin_Watts | Raymond_: When we encounter a PNG image within a file, we hold the compressed data in a buffer. | 10:39.17 |
| And we tag it with FZ_IMAGE_PNG so we know how to decompress it. | 10:39.32 |
Raymond_ | Adobe reader can't parse pdf stream, if i add a png file to a FLATE Stream | 10:39.44 |
Robin_Watts | Indeed not. | 10:39.53 |
| PNG is not supported by PDF. | 10:40.02 |
| We use it for xps and direct PNG input. | 10:40.13 |
Raymond_ | : ) , I understand | 10:40.54 |
Robin_Watts | New version of GhostDocs.apk in my casper home dir. | 14:48.41 |
| This gets underlining/strokethrough etc right, I think. | 14:49.08 |
| and paulgardiner fixed it to get the names of the converted files displayed correctly. | 14:50.24 |
| Barring accidents this is the version that mvrhel will be taking to Japan, so any smoke testing that people can do on it would be appreciated. | 14:50.51 |
| Morning mvrhel_laptop | 14:58.24 |
| New version of GhostDocs.apk in the usual place. This does underlining etc, and doesn't say "MuTmp...". | 14:58.47 |
| Hopefully this is the final version for your trip. | 14:58.55 |
mvrhel_laptop | Great. Thanks Robin_Watts | 14:59.03 |
| henrys: do you have the original J files? | 14:59.22 |
| if I recall some of them are from word, etc | 14:59.44 |
henrys | mvrhel_laptop: yes | 14:59.45 |
mvrhel_laptop | I am thinking that I should have those to test | 14:59.52 |
| showing some CJKV stuff is always good | 15:00.34 |
henrys | Robin_Watts: he doesn't seem very chatty is he upset about the deal? | 15:00.59 |
| mvrhel_laptop: the probability of those working, from what I've seen, is quite low but we can try | 15:02.23 |
| mvrhel_laptop: I should put all those files on casper though. | 15:03.04 |
mvrhel_laptop | henrys: it is worth a try | 15:03.42 |
Robin_Watts | henrys: No. Alex just isn't very chatty. | 15:04.10 |
| I suspect he's uncomfortable talking about business stuff. | 15:04.34 |
ray_laptop | morning, all. | 15:05.57 |
Robin_Watts | Morning ray | 15:06.20 |
ray_laptop | Back in the room with marginal wi-fi. We'll see if I can avoid the frequent drop-offs | 15:06.21 |
Robin_Watts | I've spotted a few lines that still appear to have problems with justification. | 15:07.15 |
ray_laptop | I played with the antennas on the access point (who knows, my boyhood dealing with rabbit ears may have taught me some antenna design) | 15:07.27 |
henrys | it looks like everyone is here which will save me sending out an email: We have purchased the Picsel outright. There will be no open source in the forseeable future (talk to me before you talk to Miles), it was my recommendation. We are sorting out how to deal with all this now. The current direction is to continue forward with paulgardiner's work but we are also going to explore maintaining the current app with outside resources. | 15:13.51 |
| So there is really no effect on Artifex engineering. | 15:13.51 |
Robin_Watts | henrys: I'm assuming therefore that we can talk about this publicly then? | 15:16.10 |
henrys | ray_laptop, mvrhel_laptop, marcosw, paulgardiner, chrisl ^^^ | 15:16.46 |
| why not? | 15:16.55 |
Robin_Watts | henrys: I didn't know if we should wait until Miles did a press release or anything. | 15:17.19 |
henrys | I don't think you should go out and make your own press announcement's but keeping it here on IRC (google search is okay) is fine. | 15:18.59 |
paulgardiner | Excellent news | 15:24.27 |
Robin_Watts | mvrhel_laptop: paulgardiner has spotted something in the current app. Just going to update to a new one. | 15:29.52 |
| From this one, it will be important to 'clear data' when upgrading versions. | 15:30.16 |
ray_laptop | marcosw: Do you want to send Søren an email with the patch for the pngalpha enhancement bug, or should I. Note that it is probably easiest to just send the new devices/gdevpng.c even though the patch is small. | 15:34.32 |
| henrys: great news! | 15:35.01 |
| marcosw: also, there seem to be two files causing segfaults that I don't see when I search the bug tracker. Are they in there? | 15:40.08 |
| tests__pdf__Bug6901014_org_chromium_ANn03F.pdf.pxlcolor.ppmraw.300.0 and tests_private__comparefiles__Bug693079.pdf.pxlcolor.ppmraw.300.0 | 15:40.10 |
| on the latter (Bug693079) I can replicate a segfault on a 32-bit windows build | 15:52.59 |
mvrhel_laptop | henrys: that is good news. let me know when and where you put the J files | 16:05.28 |
| thanks | 16:05.29 |
henrys | mvrhel_laptop: yes getting to it | 16:05.48 |
mvrhel_laptop | coming down with a cold. I hope I can nip this thing in the bud before the trip... | 16:06.21 |
Robin_Watts | paulgardiner, mvrhel_laptop: New GhostDocs.apk there now. paul: Does this solve the delay on exit for you? | 16:18.52 |
| mvrhel_laptop: When you install this version (and every version now), you need to go to settings -> applications -> Manage applications -> MuPDF and click "clear Data" | 16:19.27 |
paulgardiner | Robin_Watts: I'll give it a try. | 16:19.29 |
| Robin_Watts: so uploaded 15:37? | 16:20.56 |
Robin_Watts | yeah. | 16:22.12 |
henrys | mvrhel_laptop: in your home on casper. | 16:22.19 |
mvrhel_laptop | thanks henrys | 16:25.49 |
paulgardiner | Robin_Watts: latest version works just fine | 16:26.38 |
Robin_Watts | paulgardiner: Fab. | 16:26.46 |
ray_laptop | well, I found at least one glaring segfault. We use gcst->heap without ever initializing it | 16:45.21 |
| in ialloc_validate_spaces. Now that's only if -Z? is used, but it interferes with debugging the real segfault :-( | 16:46.45 |
| now if VS were better about single stepping :-( | 16:48.25 |
| switching to peeved and gdb (that's how desparate I am) | 16:48.47 |
henrys | chrisl:how much after greek and cyrillic is left for monotype compatibility, any ideas. assuming non technical management will want the same glyph coverage to switch? Do you have any idea? | 16:49.32 |
| chrisl: prices for greek cyrillic came back much higher than I expected. | 16:49.53 |
Robin_Watts | VS is pretty good about single stepping is it not? Macros confuse it, obviously. | 16:49.55 |
ray_laptop | Robin_Watts: I use 2008 and sometimes when I hit F10 it just runs | 16:50.49 |
Robin_Watts | Hmm. I've not had that problem. but then I use 2005 and use the button in the toolbar. | 16:51.37 |
| neither of which should really make a difference. | 16:51.49 |
mvrhel_laptop | ok. so we will avoid the j files | 16:54.28 |
| even the smart office app does not open them | 16:54.37 |
| j9 opens | 16:55.24 |
| but is missing a few things in the smart office app | 16:55.59 |
Robin_Watts | Can you open them in something else, and then save them out as .docx ? | 16:56.06 |
| maybe we'll open that instead? | 16:56.18 |
mvrhel_laptop | Robin_Watts: yes. let me try that | 16:56.20 |
Robin_Watts | I've spotted some problems with justification still. It's to do with some text coming out as a 2 byte encodings, which don't get the word spacing applied. | 16:57.14 |
| I'm trying to understand it now. | 16:57.21 |
mvrhel_laptop | ok. so I will avoid the j file with the ghostdocs apps since none of the CJKV fonts make it through the process | 17:10.05 |
| Robin_Watts: I imagine that is not a surprise? | 17:10.16 |
Robin_Watts | That some files don't load? | 17:10.33 |
mvrhel_laptop | they do work with the smart office app | 17:10.34 |
| Robin_Watts: no they load now (since I saved them to docx style) | 17:10.52 |
| the xlsx file does not load in either app | 17:11.02 |
| overall the smart office output looks ok. a couple files are missing a few things but I can make something work | 17:11.30 |
Robin_Watts | So the original file was an xlsx file, and it wouldn't load? | 17:11.41 |
| but resave it as a docx and it loads OK? | 17:11.51 |
mvrhel_laptop | Robin_Watts: sorry lots of information here | 17:12.06 |
| so there is 1 xls file | 17:12.13 |
| I converted it to xlsx | 17:12.22 |
| and it still does not load | 17:12.29 |
Robin_Watts | ok. | 17:12.37 |
mvrhel_laptop | there is one ppt file | 17:12.39 |
| that is ok. it is a little sluggish in the smart office app though | 17:12.53 |
| there are several doc files | 17:13.06 |
| I converted them all to docx | 17:13.12 |
Robin_Watts | marcosw has pointed out that there are .doc files that don't load. I have no idea why. I haven't looked. | 17:13.15 |
mvrhel_laptop | they all open | 17:13.16 |
| but one is missing a few items | 17:13.29 |
| the fonts show up in the smart office app | 17:13.39 |
| the fonts do not show up in ghostdocs | 17:13.47 |
Robin_Watts | the smart office app makes use of the fonts on the device. | 17:14.08 |
| ghostdocs does not. It only uses the inbuilt fonts. | 17:14.25 |
mvrhel_laptop | I thought we had droidsansfallback for cjkv? | 17:15.06 |
| Robin_Watts: or is that not going to work here? | 17:15.40 |
Robin_Watts | MuPDF uses that. | 17:15.56 |
| The program to convert from doc 2 pdf does not. | 17:16.06 |
mvrhel_laptop | any chance that can be made to work? | 17:16.29 |
Robin_Watts | possibly I should stick droidsansfallback in the app too. | 17:16.38 |
marcosw | ray_laptop: I had missed the pxlcolor seg faults in the regression tests, I'll open a bug after I finish the bisect. | 17:21.07 |
| ray_laptop: Can you send the pngalpha code to Søren? | 17:21.46 |
chrisl | henrys: sorry, I rushed out, forgot to change my status. I don't really have a good feel for how much over Greek/Cyrillic Monotype has. TBH, My preference would be to see what Adobe has with CPSI. | 17:23.47 |
henrys | that doesn't help us convert customers who have the ufst like the last customer you spoke with so I need to go with that. | 17:26.22 |
chrisl | henrys: Okay, I put together a list of glyphs we need over the Greek and Cyrillic ones - I should have it for you tomorrow | 17:27.46 |
| s/I put/I will put | 17:27.58 |
zeniko | henrys: a reminder: bug 694880 and bug 694906 have patches for valgrind issues awaiting review and landing | 17:30.36 |
henrys | zeniko: yes I'll get to it, I know. | 17:31.15 |
zeniko | Robin_Watts: FYI: the Freetype infinite loop issue has been fixed upstream (needed for bug 695040) | 17:31.16 |
Robin_Watts | zeniko: Ah, thanks. | 17:31.26 |
| tor8: can you handle that? | 17:31.48 |
zeniko | BTW: for when are the next releases of MuPDF and Ghostscript planned (isn't it February and August you're aiming at)? | 17:33.13 |
Robin_Watts | March/September now. | 17:33.33 |
| We've decided to slip by a month as August is always a pain cos of holidays etc. | 17:33.50 |
zeniko | good to know, thanks | 17:35.09 |
marcosw | ray_laptop: never mind, I'll send the updated pngalph to Søren. | 17:35.35 |
henrys | paulgardiner, Robin_Watts : a delivery from picsel for your perusal - on caper (Files.zip) | 17:36.08 |
paulgardiner | Oh. Christmas already. | 17:36.52 |
Robin_Watts | Hmm. No bugs database, no ATS database :( | 17:37.50 |
| Oh, Gawd. | 17:38.29 |
ray_laptop | marcosw: sorry I wasn't paying attention. I can send the email to Søren if you want. | 17:38.35 |
Robin_Watts | mvrhel_laptop: The apps I've been building for you have had the original fonts in. | 17:38.51 |
| Let me try doing a new version with the updated ones in. | 17:38.59 |
mvrhel_laptop | Robin_Watts: ok. | 17:39.02 |
henrys | Robin_Watts: if you tell me specifically what we want I will ask for it. | 17:39.03 |
Robin_Watts | henrys: The bugs database, the ATS database. Those are the two important ones. The entire testdocs repo if they have it (might have some newer stuff in) | 17:40.04 |
marcosw | ray_laptop: no problem, I already sent it. When I originally replied asking you to send it I didn't realize there was a bug report; I thought your fix was in response to an email. | 17:40.24 |
ray_laptop | marcosw: np. Thanks. | 17:40.54 |
mvrhel_laptop | bbiab | 17:41.04 |
ray_laptop | marcosw: I'm making headway on the Bug693079 problem | 17:41.37 |
Robin_Watts | mvrhel_laptop: New version there, hopefully with better fonts/ | 18:02.01 |
mvrhel_laptop | Robin_Watts: ok let me give it a try | 18:02.14 |
| thank you | 18:02.16 |
Robin_Watts | no worries. | 18:02.48 |
henrys | Robin_Watts: Miles is probably going to call you about picsel references. | 18:03.15 |
Robin_Watts | ok. | 18:03.22 |
henrys | Robin_Watts: I guess I'm confused by perpetual license: let's say a holder of this has a critical bug, do they go back to bipartite and pay? | 18:09.29 |
| or any bug | 18:09.40 |
| or paulgardiner ? | 18:09.49 |
Robin_Watts | AIUI, a perpetual licensee has bought the rights to a snapshot of the picsel code. | 18:10.00 |
| They are expected to bug fix it etc themselves. | 18:10.19 |
henrys | but if you do a binary release that borders on useless. | 18:10.21 |
Robin_Watts | It's not a binary release. | 18:10.43 |
henrys | okay the perpetuals are not binary | 18:10.55 |
| got it | 18:11.01 |
Robin_Watts | Indeed. | 18:11.03 |
mvrhel_laptop | Robin_Watts: much improved | 18:11.06 |
| thanks. Now we have CJK characters instead of little squares.... | 18:11.22 |
Robin_Watts | Perpetual Licenses are things that the liquidator sold. Basically they were flogging stuff off to make as much money as they could. | 18:11.31 |
| Before the liquidation there were no perpetual licenses, and only binary releases. | 18:11.54 |
henrys | how many of these other oems are there? Those which would require binary releases? | 18:12.17 |
Robin_Watts | BUT in order for perpetual licenses to make sense, either the licensees needed to be confident that they could handle the support burden themselves, or they needed some third party to do the support for them. | 18:12.55 |
| and bipartate was set up to be that third party. | 18:13.06 |
| no perpetual licensee is required to use bipartate, but the expectation is that many will. | 18:13.28 |
| How many OEMS? How many perpetual licensees? No firm idea on either count. | 18:13.51 |
| I'd assume that Miles knows how many OEMs. | 18:14.10 |
henrys | I asked him first | 18:14.23 |
Robin_Watts | I'd expect of the order of 3-5 perpetual licensees, but honestly that's pretty much guesswork. | 18:14.27 |
henrys | I hope that monthly app number is real and not a "record month", as I suspect it might be. | 18:15.06 |
Robin_Watts | henrys: likewise. | 18:15.20 |
| If we can recoup what we have paid within a year or so, from app sales alone, I think we can be happy. | 18:16.33 |
| I was rather hoping that miles would get a list of the OEMs and past OEM agreements as part of the sale. | 18:17.08 |
henrys | apparently there is terabytes of information we'll be downloading and paper work in Glasgow to retrieve. | 18:18.22 |
Robin_Watts | henrys: That makes sense. | 18:18.50 |
| The testdocs from 2009 was 500GB alone. | 18:19.21 |
henrys | I do believe he went in thinking the oem market could be 0 and the app market + code was worth it. | 18:19.21 |
| business decision wise. He's more a gambler than I. | 18:19.55 |
Robin_Watts | The thing that convinced me was the idea that at worst, we should get our money back from app store sales within a year. | 18:20.09 |
| any OEM sales are a bonus. | 18:20.20 |
henrys | Robin_Watts: sure if I saw a documented trail of monthly payments, but I just don't know exactly what that monthly number really is. | 18:21.43 |
| mean and median could be quite different ;-( | 18:22.19 |
Robin_Watts | I was leaving that to Miles. You'd assume that the liquidators would have offered some sort of support for their claims. | 18:22.44 |
henrys | Robin_Watts: my impression was he couldn't get a straight answer, but I might have missed something. | 18:23.44 |
marcosw | Robin_Watts: is there anything else I should be doing at this point for GhostDocs word testing? If not I'm going to run comparisons for the powerpoint files. | 18:42.41 |
Robin_Watts | marcosw: Sorry, was on skype with Miles | 18:53.02 |
| marcosw: Urm... not for now. | 18:53.48 |
| I've fixed the underlines/strikeouts. | 18:54.02 |
| but it's not worth rerunning just for that. | 18:54.12 |
| Maybe we could rerun if I get the last few justification cases working. | 18:54.50 |
| but for now, ppt is reasonable to look at. | 18:55.00 |
marcosw | Robin_Watts: okay | 18:55.15 |
| Robin_Watts: can I ask a (probably) stupid question? Why did the underlines/strikeouts need fixing? That seems like the kind of thing that shouldn't have been broken to begin with. | 19:02.50 |
Robin_Watts | marcosw: They work in the Smart Office app. | 19:03.28 |
| They don't work in the pdf export. | 19:03.33 |
| or rather, they didn't. | 19:03.37 |
| The pdf export code was written (about a year or two ago - since I left), but never really used at Picsel. | 19:04.12 |
| so there were teething troubles. That's what I've been fixing to make it work better in our demo. | 19:04.32 |
henrys | bbiab | 19:59.30 |
| bbiaw | 20:38.31 |
| Forward 1 day (to 2014/02/20)>>> | |