| <<<Back 1 day (to 2011/08/24) | 2011/08/25 |
henrys | mvrhel2:I asked alexcher to test ray_laptop's solution I think another person should go through the setup and running before we send. | 04:54.08 |
alexcher | henrys: I'm doing it now. | 04:54.29 |
mvrhel2 | sounds like a good idea | 04:54.35 |
| I was not sure from Ray's email if he was asking me or alex to send it to the customer | 04:54.54 |
henrys | well it really should be marcos but he's probably not available now so I guess I'll do it once it is tested. | 04:56.13 |
alexcher | ray_laptop: Ray's program doesn't consider a case when a sample file sets some random resolution. | 04:56.53 |
henrys | can you fix it and post it back to the email list so ray_laptop can approve the change to his code? | 05:02.44 |
| sorry if I'm stating the obvious | 05:03.48 |
alexcher | henrys: sure. | 05:06.28 |
mvrhel2 | good night all | 07:06.52 |
kens | Hi Robin_Watts | 08:15.32 |
Robin_Watts | Morning | 08:15.52 |
kens | How is the jet lag today ? | 08:15.59 |
Robin_Watts | OK so far, but ask me again this evening. | 08:17.06 |
kens | :-) | 08:17.10 |
| Nice to see someone else in here, chrisl has been out for the last 3 days (but back today), its been very quiet... | 08:17.33 |
Robin_Watts | I suspect today will be largely slaving through email etc. | 08:19.45 |
| Just rebooted the PC umpteen times as windows updates etc. | 08:20.04 |
| Time to go for a run. This could be painful. | 08:20.32 |
kens | Yes, I expect so.... | 08:20.40 |
chrisl | 'morning guys | 08:25.28 |
kens | Hi chrisl , good break ? | 08:25.36 |
chrisl | Yes, until I got back - seems my cat isn't too well | 08:25.57 |
kens | Sorry to hear that. You parents looking after her ? | 08:26.19 |
Robin_Watts | morning chrisl. poor mavis. | 08:26.49 |
chrisl | hi Robin_Watts, I take it the holiday was good? | 08:27.15 |
Robin_Watts | chrisl: Fab thanks. | 08:27.36 |
| But glad to be back to weather where I'm not melting all the time. | 08:27.52 |
chrisl | You prefer drowning, then? It's chucking it down here | 08:28.30 |
kens | Dry here | 08:28.42 |
chrisl | kens: yes, my parents reckoned she wasn't eating enough, and she seems to be a bit under weight - just waiting until I can call the vet for an appointment :-( | 08:28.49 |
kens | :-( | 08:29.01 |
chrisl | The thing is, she seems happy, alert - sitting between me and the keyboard, purring away..... don't quite know what to make of it | 08:29.41 |
Robin_Watts | chrisl: Maybe she's just been shulking at you being away. | 08:30.36 |
chrisl | Robin_Watts: that might be part of it, but it wouldn't explain the weight loss. | 08:31.22 |
| I suspect it's teeth or gums that are the problem - or maybe a deeply lodged fur-ball | 08:32.28 |
kens | Not eating, pining. | 08:32.32 |
chrisl | It's time she had a checkup anyway | 08:32.43 |
| Although, by the look she just gave me, she can even tell when I *type* vet! | 08:33.11 |
| Much to her displeasure (it seems), she is booked in to the vets tomorrow 11:20 am - and she's just gone and hidden under the desk....... | 08:38.38 |
kens | She's got a good fgrasp of English | 08:38.54 |
chrisl | I think she has a good "feel" for what's going on around her | 08:39.31 |
kens | :-) | 08:39.43 |
chrisl | kens: So, what's going on with the cidfmap CID font thing - seems to have been some confusion? | 08:40.01 |
kens | Lots of confusion. | 08:40.10 |
| Just a mo | 08:40.13 |
chrisl | Well, it seems to be assigned to you now, which seems wrong in every way....... | 08:40.38 |
kens | It is wrong. I opened it but it was assigned toyou before. | 08:41.01 |
| Anywya: | 08:41.05 |
| Marcos got mail from the customer 'doesn't work' and 'Ken opened a bug for it' | 08:41.24 |
| So he changed hte enhancement to a customer bug. | 08:41.36 |
| I change dit back and said 'we told them hot to do it' | 08:41.49 |
| how | 08:41.52 |
| Ray tried it and it didn't work for him. | 08:42.00 |
| Long discussion on IRC ensued. | 08:42.11 |
| Turns out if the path contains the string 'Resource' then it sets GenericResourceDir, and all hell breaks loose. | 08:42.32 |
chrisl | Yeh, I read most of it - I was still a bit confused | 08:42.42 |
kens | Of course, my path didn't contain that, Ray's did. | 08:42.46 |
chrisl | So, is that a bug in itself? I seems wrong to me | 08:43.06 |
kens | Customer is still complaining it doesn't work. But they are assigning GenericResourceDir, which I didn't do. | 08:43.17 |
| I think that it is a bug, I believe Ray said so. | 08:43.32 |
| No tracker opened though I don't think. | 08:43.42 |
| Was more mail from the cuistomer last night. | 08:44.01 |
| I was hoping to leave it to Ray. | 08:44.11 |
| However, the original enhancement should still be an enhancement, with no customer priority. | 08:44.37 |
| It *should* be assigned to you, I gess Marcos re-assigned it to me and I ddn't notice. | 08:44.55 |
| In the confusion. | 08:45.03 |
chrisl | Ray assigned it to you, I believe, I've taken it back now | 08:45.15 |
kens | OK, thanks. | 08:45.22 |
| *now* its bucketing down | 08:45.29 |
| I guess Melanie is getting wet. | 08:45.40 |
chrisl | Oh, it's dry here, now ;-) | 08:45.50 |
kens | It looks to me like the customer is using the API and the DLL, not the GS test harness but I think they ought to behave the same. Feel free to try it out if you like. | 08:46.29 |
| But I don't believe the customer's problem is really anything to do with the bug, that's their mis-understanding | 08:46.51 |
chrisl | It sounds like you and Ray have covered bases - I think the *bug* is treating any path with "resource" in it as the GenericResourceDir - that seems utterly wrong | 08:47.38 |
kens | Yes, agree. | 08:47.49 |
| I notice in the customer's log that they get an error about GenericResourceDir not pointing to a valid directory | 08:48.13 |
chrisl | Which, I'm afraid, is really going to need Alex to look at - so much for helping to reduce his tally..... | 08:48.50 |
kens | Ah, and htey fail to find the CMap which is exactly the problem Ray encountered. | 08:49.00 |
| Well, I have knocked 4 of his list so far. | 08:49.14 |
| Sadly, as fast as I remove them, new ones come in. | 08:49.26 |
| I think I'm going to have to stop looking at them for a few days too, my own list is growing. THough not of customer issues. | 08:49.59 |
| You know, the cusomter has a really weird CMap in there. | 08:50.25 |
| UniJis-UCS2-HW-H | 08:50.40 |
| Well, we do ship it. | 08:51.24 |
chrisl | Not a widely used CMap, but part of the Adobe set | 08:52.23 |
| Looking at the customer's mail, they give two paths with "Resource" in them, | 08:52.29 |
kens | Hmm, its date is 2-=04-2011, I wonder if the problem is that they are using 9.01 and its not in there | 08:52.35 |
| Both of which are going to fail. They *have* been told not to do that..... | 08:53.09 |
| Can Git tell me when a file was added ? | 08:53.23 |
kens | goes to try | 08:53.37 |
chrisl | probably git log <filename> | 08:54.03 |
| It was in 9.01 | 08:54.47 |
kens | OK, I was having capitalisation trouble. | 08:55.00 |
| I htink the -I./Resource is re-assigning the GenericResourceDir, which is why they get that warning | 08:55.21 |
| And that causes the CMap error, just as Ray found. | 08:55.33 |
chrisl | Yeh, that is probably the problem. | 08:56.20 |
kens | I think so, I'm writing a mail. | 08:56.38 |
chrisl | Okay, I think I'll get cracking on doing the enhancement - sounds like we'll actually need it at some point...... | 08:58.03 |
kens | I think it would be useful anyway, but the GenericResourceDir being set from -I is bad. | 08:58.48 |
chrisl | I don't see immediately where that's happening, but I'll have a quick look later | 09:00.45 |
kens | OK, no rush but it would be nice. | 09:00.58 |
chrisl | I think I'll get another coffee now..... | 09:01.04 |
kens | Me too.... | 09:01.32 |
Robin_Watts | ow. | 09:31.07 |
chrisl | Robin_Watts: general "ow", or a specific "ow"? | 09:31.59 |
Robin_Watts | running after 2 weeks off and a stone gained. ow. | 09:32.22 |
chrisl | So not a new or recurred injury, then | 09:32.56 |
Robin_Watts | no. | 09:33.26 |
chrisl | Well, that's good | 09:33.53 |
kens | I think I'm reaching the end of any thing I can do to help Alex. | 09:34.46 |
Robin_Watts | ok, that's email caught up I think. | 12:55.45 |
kens | Not bad. | 12:56.45 |
| Robin_Watts : did you see that bug just came in ? Cross-compiling | 12:58.37 |
Robin_Watts | no... | 12:58.48 |
| mail is refetching now. Maybe I hadn't got everything in before. | 12:59.09 |
kens | #692463 | 12:59.11 |
| doesn't say the target though | 12:59.21 |
Robin_Watts | He's building 9.00 without using any of our makefiles. | 13:04.09 |
| and finding it doesn't work. | 13:04.16 |
kens | Surprise! :-) | 13:04.24 |
Robin_Watts | mmm. | 13:04.29 |
kens | Well, as long as he doesn't expect us to fix it :-) | 13:04.48 |
Robin_Watts | kens: That wasn't a customer bug, right? :) | 13:10.12 |
kens | Nope. | 13:10.17 |
Robin_Watts | Phew :) | 13:10.21 |
kens | Just saw it in passing, thought of you | 13:10.25 |
Robin_Watts | I just posted a comment saying we probably wouldn't be looking into it. | 13:10.45 |
kens | Seems reasonable | 13:10.52 |
Robin_Watts | Gah. They sent me a board with no PSU and no RAM. | 13:32.09 |
| and it's only tested out with a limited number of RAM module types. And the sales link they sent is in the US :( | 13:32.32 |
kens | Oh, oh.... | 13:32.44 |
Robin_Watts | The link for the american site has it costing $102.94. | 13:40.51 |
kens | Yow! | 13:41.07 |
Robin_Watts | Want to guess how much a seemingly identical module costs in the UK ? | 13:41.13 |
kens | get it sent to Miles ? | 13:41.14 |
| for next meeting | 13:41.19 |
Robin_Watts | 11.04 inc VAT. | 13:41.37 |
kens | weird | 13:41.51 |
| not the usual way | 13:42.17 |
henrys | Robin_Watts:I wouldn't slow down the project for that. | 14:17.10 |
| kens:do we know how 689933 was fixed? 1.6G -> 50M | 14:21.58 |
kens | No idea, sorry, its Alex's comment | 14:22.14 |
| suspect transparency improvement | 14:22.37 |
Robin_Watts | henrys: Slow down the project? | 14:32.25 |
| Currently, until I get a PSU, some RAM and the tools I can't even start. | 14:32.44 |
| Getting RAM that is known to work is the stumbling block at the moment. | 14:33.09 |
| I'm tempted to just order the RAM that I can get and see if it works; if it doesn't, I'll have to get one of you to order the specific RAM module he mentions and bring it back with me from the staff meeting. | 14:34.24 |
henrys | I meant spending $100 and doing the project sooner is best. | 14:35.44 |
| why can't we order the ram and fedex it to you? | 14:37.38 |
Robin_Watts | henrys: You could. | 14:37.48 |
kens | Would be quicker thna waiting till the meeting | 14:38.01 |
Robin_Watts | Great, let's do that. | 14:38.08 |
kens | How much would it cost to buy some and see ? | 14:38.13 |
henrys | well that's an option I can do let me know. | 14:38.33 |
Robin_Watts | kens: 20 quid or so (11.04 + postage). | 14:38.36 |
kens | Hmm, well it ouwld be cheaper | 14:39.07 |
Robin_Watts | kens: Allowing for my time googling around trying to find an equivalent one, and the problems that could be caused by having a misbehaving system, it's probably best to just go ahead and order that particular one. | 14:41.50 |
kens | ok | 14:42.01 |
Robin_Watts | henrys: Could you order that/ship it to me please? | 14:42.03 |
kens | whatever is best | 14:42.05 |
henrys | yes was I on that email? | 14:45.08 |
Robin_Watts | Yes. 19th August. | 14:45.28 |
| Subject: RE: Additional Tools | 14:45.44 |
henrys | so just use his memory4less link? | 14:50.04 |
kens | well they say that's known to work | 14:50.20 |
henrys | okay I'll order this, anything else? | 14:51.25 |
| do you want me to get the psu also? | 14:56.15 |
| I'll just do 3 day shipping on this. | 14:57.09 |
Robin_Watts | henrys: I can get a PSU. | 15:00.58 |
| Either I'll rip one out of an old PC, or I'll use my spare one here. | 15:01.18 |
| How can they need 450W for an SBC? | 15:01.31 |
| The beagleboard runs from a USB socket :) | 15:01.52 |
kens | Shuldn't think they cuold disipate the heat if it needed that | 15:02.07 |
Robin_Watts | PC PSUs provide different percentages of the power on the different lines; maybe they are only using the 5V one or something, and so need a larger PSU to ensure that enough is available on that. Or something. | 15:03.11 |
kens | Hmm, maybe..... | 15:03.27 |
henrys | okay I ordered it and will fedex it when it gets here. | 15:07.31 |
Robin_Watts | henrys: Thanks. | 15:08.16 |
| henrys: Did you try the memory squeezing stuff ? | 15:08.34 |
henrys | yes I did and now I'm procratinating fixing what it has found. | 15:15.54 |
Robin_Watts | does the mvrhel2 summoning dance. | 15:46.08 |
| Hey ray. How was the cruise? | 15:49.14 |
ray_laptop | Robin_Watts: the whole vacation was fun, but lots of work to keep 'doing' stuff. The kids really enjoyed all the new experiences, however. | 15:50.37 |
| for most of it, there were 11 of us (my mom and dad, three sisters and one brother-in-law and the 5 of us) | 15:51.36 |
Robin_Watts | I can imagine that it's significant work to keep kids entertained. It's bad enough travelling with Helen :) | 15:52.33 |
ray_laptop | we missed the sea turtle egg laying and hatching, however. The bugs were horrible and we couldn't stand to stay out waiting for them | 15:52.35 |
| I had fun on all the rides with my kids and in the ocean. My kids really liked the dolphin experience in the Bahamas, too | 15:54.02 |
Robin_Watts | We saw the sea turtles laying in Costa Rica. Had to go onto the beach silently in the middle of the night in the pitch dark. | 15:54.30 |
| Dolphin experience? Sounds cool. Never done anything with Dolphins. | 15:55.04 |
ray_laptop | this was in Florida -- Melbourne Beach. There's a turtle preserve there and they come right up to the motel patio (they like laying right along the wall) | 15:55.35 |
henrys | ray_laptop:I did the dolphin thing when my kids were little, it was a lot of fun. | 15:57.52 |
ray_laptop | I'd rather be with dolphins in open water, but some of the 11 couldn't (or didn't want to) do that, so this was more just getting to touch and a few trained behaviors (hug, kiss, dance) instead of a swim with them | 15:57.53 |
| the age limit was 6 so Austen wouldn't have able to do the swim, but he really wanted to -- he was in a flotation vest and get swimming out deeper to be with it. | 15:59.19 |
| I had to hang on to his strap like a puppy on a leash :-) | 15:59.48 |
henrys | ray_laptop:did you want to have the meeting today? I don't see much to talk about | 16:00.45 |
ray_laptop | henrys: I saw that alexcher was going to add a 'default' case if the resolution wasn't available. For this customer it isn't needed, and I figured the setpagedevice would just get the error and then it woul 'revert' as per spec. | 16:02.36 |
henrys | I think he just sent out email about that. | 16:03.39 |
kens | I just got it | 16:04.27 |
| ray_laptop : re bug#692372 the handle leaking, did you have a fix for that ? | 16:09.10 |
| ray_laptop : ? | 16:11.07 |
ray_laptop | henrys: I just saw alexcher's email and replied. | 16:12.07 |
| kens: well, yes and no. First, I am not able to reproduce the handle leakage, even with the customer's front end. | 16:12.46 |
kens | Oh, I can. | 16:12.54 |
| If you send me the patch I can test it. | 16:13.04 |
ray_laptop | kens: my patch to add 'finalization' to the imager_state so that we can rc_decrement the ICC stuff properly, worked fine on Windows and does the proper thing, but I have a segv on linux :-( | 16:14.15 |
kens | Oh, that's not great. | 16:14.29 |
| The only reason I asked is the customer sent mail to me while you were away. | 16:14.43 |
henrys | kens, ray_laptop:as I understand that you would have tear down all allocations not to leak. | 16:14.51 |
ray_laptop | my priority for today is to get on xxgdb or ddd on peeves and find out why it's bombing | 16:15.06 |
henrys | assuming you are creating a new instance each time. | 16:15.12 |
kens | Its specificlaly the semaphores henrys | 16:15.13 |
| OK ray thanks. | 16:15.24 |
| If you need anything testing I'll be happy to do it. | 16:15.34 |
henrys | kens:how can you delete the instance and still have access to pointer inside the instance - doesn't make sense. | 16:15.54 |
kens | What instance henrys ? | 16:16.09 |
henrys | in particular the entire allocator would leak. | 16:16.13 |
kens | The graphics state wasn't being reference counted to 0 | 16:16.18 |
ray_laptop | henrys: it doesn't leak memory, but the issue is that in the 'alloc_restore_all' at the end, we were just freeing the graphics_state/imager_state stuff in various save levels and not freeing the ICC stuff or the closing the semaphores as a result | 16:16.32 |
henrys | the bug says you are deleting the instance. | 16:16.37 |
kens | Yes, but it doesn't clear the gstate | 16:16.47 |
ray_laptop | kens: I don't ref count the g/i states, but when one _does_ get freed, then the icc rc are kept straight by the finalization. That counting works fine | 16:17.22 |
henrys | I have a meeting bbiab | 16:17.35 |
kens | OK fair enough, I didn't know hte exact details ;-) | 16:17.42 |
henrys | support meeting canceled | 16:18.14 |
ray_laptop | kens: I had discussed this with mvrhel, and this seemed reasonable | 16:18.17 |
| henrys: OK, thanks | 16:18.22 |
kens | Yes. I kind of lost touch around then, because it wasn't my field ;-) | 16:18.39 |
ray_laptop | henrys: I was thinking that Michael would send the info to the customer since he's been the primary point of contact on the issue (converting color...) | 16:19.30 |
| henrys: but if you want me to, let me know. | 16:19.49 |
henrys | michael wanted to get cut out of it, so I was going to do it. If you want to do it have at it. It supposed to marcos but he's on vacation. | 16:20.29 |
kens | OK I'm heading off now. I'l lbe out in the morning, we moved riding from Wednesday to Friday this week. Have a goood day everyone. | 16:21.13 |
ray_laptop | henrys: Since I did most of this, then I'll step in and forward it. That way if they have questions/issues he can contact me | 16:22.25 |
henrys | sounds good | 16:22.40 |
| generally I'd really like to get the halftone stuff working with devices so it will work with all the languages and the mupdf parser once it's ported to the gs library. | 16:29.59 |
henrys | thinks we have to stop solving problems with a postscript interpreter. | 16:30.29 |
ray_laptop | henrys: that's fine in the longer term. The issue of replacing 'setpagedevice' functionality is bigger since we need to map out the way parsers should do the things normally done by gs_init and other .ps | 16:33.02 |
marcosw_ | henrys: sorry I'm late for the support meeting. | 16:33.53 |
henrys | it's canceled | 16:34.06 |
Robin_Watts | hey marcosw_: Have fun in spa. | 16:34.10 |
| Which grandstand are you in ? | 16:34.22 |
marcosw_ | henrys: damn, it's the high point of my week. | 16:34.28 |
henrys | marcosw_:we were sort of eager to get that plank device in maybe that can be your high point | 16:34.42 |
| ray_laptop:I'll add that to the meeting agenda | 16:35.37 |
marcosw_ | Robin_Watts: I'm in Gold 3, which is the covered Eau Rouge grandstand. I'm in Row 1, Seat 025, so look for me on television. | 16:35.51 |
Robin_Watts | marcosw_: I will :) | 16:36.03 |
ray_laptop | some of the issues besides halftone selection are: profile selection, N-up, BeginPage / EndPage, initial matrix, tray/media selection. I'll try to put together a list of things people do with PS and/or setpagedevice now. | 16:36.09 |
| marcosw_: cool :-) | 16:36.24 |
| marcosw_: I was on vacation -- what event is it ? | 16:37.13 |
Robin_Watts | Formula 1 Belgian Grand Prix | 16:37.41 |
| One of the best tracks on the calendar. | 16:38.07 |
marcosw_ | henrys: I started working on plank support last night. For bmpcmp we'll have to convert the CMYK data to RGB, the question is should we do so such that each plane is a separate gray scale file? | 16:38.20 |
| Also, what's the difference between plank and psdcmyk? | 16:39.27 |
| Robin_Watts: "one of" ? | 16:39.35 |
henrys | plank is planar? | 16:39.44 |
Robin_Watts | marcosw_: bmpcmp can handle cmyk. | 16:40.01 |
henrys | so the output of plank is chunky so this should be the same as the other devices | 16:40.39 |
Robin_Watts | (that is, it will compare the c, m, y, k separately, but the output 'diffs' will be in rgb, if memory serves) | 16:40.40 |
henrys | right? | 16:40.42 |
Robin_Watts | henrys: yes. | 16:40.48 |
| marcosw_: plank runs internally all planar, and just converts to chunky on output. | 16:41.08 |
henrys | adding the device will let mvrhel test his thresholding changes but it doesn't really test the plank device, for which we need a plan. | 16:42.08 |
| no pun intended. | 16:42.34 |
Robin_Watts | To test planb we want to run pbmraw and diff it to planb. | 16:43.55 |
| To test plang we want to run pgmraw and diff it to plang. | 16:44.26 |
marcosw_ | Is there an easy way to add the plank device during the ./configure step? "--with-drivers=plank" removes all the other devices. Or should I just make it a device that's enabled by default? | 16:44.42 |
Robin_Watts | It's enabled on windows by default. | 16:44.57 |
| So having it enabled on unix wouldn't hurt. | 16:45.09 |
marcosw_ | okay, I'll enable it under Linux as well. | 16:45.32 |
Robin_Watts | Hmm. not planb, planm. | 16:45.50 |
| To test plan we diff it against ppmraw | 16:46.12 |
henrys | the cmyk device is the issue we discussed before and you thought we would need a fuzzy compare or something? | 16:47.29 |
Robin_Watts | pamcmyk4 and pamcmyk32 are the equivalents to plank and planc. | 16:47.29 |
| henrys: I'd be tempted to run a compare and see. | 16:48.10 |
| bmpcmp can be invoked with -t 16 say. | 16:48.35 |
| but that will only work if there are fewer than 1000 diffs. | 16:48.51 |
marcosw_ | So we want to do a bitmap comparison from plank to pamcmyk4? | 16:49.02 |
Robin_Watts | Yes. | 16:49.36 |
| (or at least, we want the ability to do it - at the moment we know it won't match, until we get the pattern stuff sorted) | 16:50.09 |
| I suspect it's something we're going to want to run several times. | 16:50.26 |
marcosw_ | It's pretty easy for me to run it locally. | 16:51.05 |
ray_laptop | Robin_Watts: while it is slow, I thought Michael's temp patch to force clist patterns was to allow us to do initial testing. | 16:51.38 |
marcosw_ | While I'm adding default devices, should I add the other plan* and pam* devices as well? | 16:51.41 |
Robin_Watts | ray_laptop: Oh, yes, I forgot I had to disable that earlier :) | 16:52.03 |
henrys | marcosw_:can you run it locally before going live with it? I am worried the clist pattern stuff is going to kill the cluster for some files. | 16:54.30 |
marcosw_ | henrys: sure, I can do that. Should I open a bug for files that are particularly slow? | 16:56.15 |
| BTW, do we want do test plank at 72 and 300 dpi? And just direct from file to bitmap (i.e. not via ps/pdf/pxlmono/pxlcolor)? | 16:57.12 |
Robin_Watts | marcosw_: That sounds reasonable to me. | 17:01.58 |
| at least initially. | 17:02.19 |
henrys | out of meeting | 17:10.40 |
Robin_Watts | Are we expecting mvrhel2 today? | 17:11.16 |
henrys | I just want to know if we can reasonably run the cluster with the devices. | 17:11.58 |
| Robin_Watts:I thought he'd be in. | 17:12.44 |
| well the pcl tests do 600 and ps 300 the way it is set up. | 17:14.16 |
| ^^^ marcosw_ | 17:15.14 |
marcosw_ | henrys: right, I forgot about plank in GhostPCL. | 17:19.44 |
henrys | errand back in an hour | 17:20.53 |
Robin_Watts | Aha. An mvrhel2. | 17:25.39 |
| mvrhel2: Got time to talk about this planar pattern stuff? | 17:25.55 |
mvrhel2 | hi Robin_Watts | 17:26.03 |
| sure | 17:26.08 |
Robin_Watts | I added a copy_plane thing to gxclip2.c, but it never gets called. | 17:26.45 |
mvrhel2 | oh well yes | 17:26.52 |
| we need to add in some other changes to make it be used | 17:27.06 |
| can you commit that and I can go from there | 17:27.19 |
Robin_Watts | Oh. ok. | 17:27.33 |
mvrhel2 | that was my bottleneck I belive | 17:27.41 |
Robin_Watts | If you're going to take over, that's grand. | 17:27.46 |
mvrhel2 | well, I may drag you back in though | 17:28.02 |
| very likely | 17:28.04 |
Robin_Watts | Sure. | 17:28.06 |
| It's completely trivial. | 17:28.35 |
mvrhel2 | also, the current code wont use it due to the fact that for planar devices all patterns are now stored in a clist | 17:28.43 |
Robin_Watts | Just a copy of copy_color with an extra plane arg. | 17:28.53 |
| yes, I disabled that locally. | 17:29.00 |
mvrhel2 | ok | 17:29.03 |
| good | 17:29.05 |
| so henrys and I were talking that you had a way to compare the plank output with pkmraw in a fuzzy way so that we can check if we have any issues? | 17:30.02 |
| that is comparing the plank device and the pkmraw device | 17:30.22 |
| output | 17:30.25 |
Robin_Watts | mvrhel2: We've just been talking about that. | 17:31.04 |
mvrhel2 | oh | 17:31.11 |
| need to read the logs | 17:31.15 |
Robin_Watts | I have no magic bullet, but I think comparing and then looking using bmpcmp -t 16 will be a start. | 17:32.01 |
| mvrhel2: committed. | 17:32.18 |
mvrhel2 | Robin_Wattts: ok thanks | 17:32.28 |
Robin_Watts | mvrhel2: I can see where we spot that the output device is planar and set the memory buffers to be planar. | 17:32.55 |
mvrhel2 | yes. that all works fine | 17:33.14 |
Robin_Watts | and I can see where we need to do something different on the painting side, I think. | 17:33.21 |
mvrhel2 | the pattern itself is nice planar data | 17:33.24 |
Robin_Watts | line 345 of gxp1fill.c | 17:34.01 |
| rather than passing in 'tile_colored_fill' I suspect we need to pass in 'tile_planar_colored_fill' or something. | 17:34.21 |
| maybe it's just my jetlagged brain, but I just can't see at that stage how we can tell it's planar data. | 17:34.55 |
mvrhel2 | I have that all figured out I think | 17:35.07 |
| it was just the clip2 device that was puzzling me | 17:35.27 |
| need to clean up some things I was working on in color then I am going to get on this now | 17:36.05 |
Robin_Watts | Now you have me worried that I've overlooked something in the clip2 device :) | 17:36.08 |
mvrhel2 | hehe. probably not | 17:36.14 |
ray_laptop | well at least one person (William Bader) is reading our docs :-/ | 17:36.19 |
mvrhel2 | just me thinking there was some issue that I had to worry about | 17:36.36 |
| Robin_Watts: ok, so we made it in tile_clip_copy_plane | 17:50.34 |
| now let me see what explodes next... | 17:50.50 |
| ok that seems to be working | 17:51.12 |
Robin_Watts | really? Nice one. | 17:51.26 |
mvrhel2 | hey it ran all the way through | 17:51.41 |
| lets see what the output looks like | 17:51.51 |
Robin_Watts | but what does it look like :) | 17:51.53 |
mvrhel2 | very close | 17:52.39 |
| looks like we are copying the same plane over | 17:52.49 |
| probably my goof | 17:52.56 |
Robin_Watts | Can I see your code changes? | 17:53.01 |
mvrhel2 | this should be easy to track down | 17:53.03 |
| hold on let me check this first | 17:53.11 |
| ok. I found the issue. it is my failure to move my buffer ptr | 17:56.00 |
| let me fix this then I will let you review it | 17:56.12 |
Robin_Watts | goes to help cook. jetlag is kicking in now. will check back in a bit unless I fall asleep. | 18:03.41 |
mvrhel2 | no problem. thanks for your help Robin_Watts | 18:04.02 |
Robin_Watts | np. I'll be back to looking at the color rounding stuff tomorrow. | 18:04.36 |
| mostly to try and find a nice way to get the code both fast and reusable in many places. | 18:04.53 |
mvrhel2 | Robin_Watts: are you still here? | 18:12.38 |
| sorry nevermind | 18:12.51 |
| please go cook | 18:12.54 |
henrys | mvrhel2:don't forget about the keyword CLUSTER_UNTESTED | 18:19.20 |
mvrhel2 | dumb cable modem had a seizure | 18:21.38 |
| henrys: you mean when I commit tthese planar changes? | 18:22.17 |
henrys | if a change can't have an effect, I thought that was the case for your last change. | 18:25.11 |
| no big deal just wasn't sure you knew about it. | 18:25.27 |
mvrhel2 | hmm yes I should have included it in that last commit | 18:25.44 |
henrys | are irclogs archived? | 20:58.58 |
mvrhel2 | I dont know if they are | 21:13.53 |
| I think ray_laptop has a log some place | 21:14.03 |
| an archive I mean | 21:14.11 |
| so I seem to have the non-clist case working now for planar device with patterns | 21:14.50 |
| I had one hiccup that I am testing now | 21:15.03 |
| heading out for a bit. back shortly | 21:16.37 |
henrys | okay | 21:16.43 |
mvrhel2 | if the cluster test comes back ok then I will commit and re-enable the non-clist case | 21:17.22 |
| then we need to do the plank vs pkmraw comparison | 21:17.37 |
| with fuzzy | 21:17.54 |
henrys | mvrhel2:did you want someone in particular to look at 692448 | 21:19.59 |
| ? | 21:20.00 |
Robin_Watts | mvrhel2: patch looks good to me, FWIW. | 22:51.19 |
mvrhel2 | which patch? | 23:10.57 |
| oh did you see the clusterpush | 23:11.09 |
| Robin_Watts | 23:11.12 |
| ok. I am going to go ahead and push this now then | 23:11.48 |
| or is a second | 23:11.53 |
| oh henrys | 23:13.07 |
| 692448 should be closed | 23:13.12 |
| this was from when the tiff device was broken | 23:13.25 |
| that was what the issue was | 23:13.28 |
| :) | 23:13.31 |
| I will go ahead and do it | 23:13.47 |
| ok. that is done | 23:21.10 |
| Robin_Watts and henrys: I was wondering if there was some way to easily fake out the cluster so that it ran the plank device when it thought is was doing the pkmraw device (via a code change) and this would allow the cluster to point out any diffs that we can look at with bmpcmp | 23:22.20 |
| I would expect plank and pkmraw to be identical though with the fast thresholding turned off | 23:22.47 |
| that would at least get us through that check | 23:23.00 |
| brb | 23:23.02 |
| Forward 1 day (to 2011/08/26)>>> | |