| <<<Back 1 day (to 2011/08/22) | 2011/08/23 |
robin_watts_mac | heads for Bangkok. bbl, hopefully. | 10:42.40 |
ManDay | tor8: You got a sitrep on your planned reader? any time yet? | 13:19.25 |
tor8 | ManDay: nope :( | 13:21.04 |
ManDay | ok thanks for the feedback (i know im a pain in the arse, forgive me :>) | 13:21.57 |
| dont forget to look into zathura before you start though | 13:22.20 |
| it's slowly getting there | 13:22.29 |
ray_laptop | morning, all | 14:22.52 |
kens | Hi Ray | 14:23.43 |
ray_laptop | kens: 692300 is still a problem for customer 780. I'm writing up what I learned on the phone call from them yesterday. | 14:34.26 |
kens | OK, that would be good. All the problems I know about are resolved, so new information is useful. | 14:34.56 |
ray_laptop | my brain is fuinctioning at about 90% today (up from about 50% yesterday). My cold is almost gone | 14:35.00 |
| kens: basically, the -I switch doesn't help with CIDFonts | 14:35.35 |
kens | It does for me. | 14:35.50 |
| But I'll investigate again when I see you r findings. | 14:36.27 |
ray_laptop | kens: it 'might' be working if the directory pointed to by the -I has a name 'CIDFont' IFF it is the first path on the list. Then it has the side effect of setting the default GenericResourceDir | 14:48.35 |
kens | Hmm, I found that any directory on the search path which had a sub-directory called CIDFont would work, but I did not test it extensively. | 14:49.13 |
| Apparently the resoruce code tests each directory in the search path for specifically named sub-directories by resource name). | 14:49.48 |
| But like I said, I didn't test exhaustively. | 14:50.07 |
ray_laptop | kens: not what I found. For instance, when I set -Ic:/Resource/CIDFont then it couldn't find the CMap files even though they are in %rom%Resource/CMap | 14:57.08 |
| kens: and if the CIDFonts are in a directory c:/Resource/CIDFonts then it didn't work at alll | 14:57.47 |
kens | Well it wouldn't find the CMaps if you specified the CIDFont directory | 14:58.05 |
| What you do is specify a folder called (eg) MyCIDFonts, which has a sub-direcotry CIDFONt. In /MyCIDFonts you put the cidfmap. | 14:58.32 |
ray_laptop | kens: because it only assigned the default GenericResourceDir and only concatenated CIDFont/ | 14:58.34 |
kens | Then you say -I/MyCIDFonts | 14:58.45 |
| THat worked for me | 14:58.50 |
| That picks up the cidmap, and then the CIDFonts relative to it. | 14:59.21 |
ray_laptop | kens: did you see the warning about an incorrect GenericResourceDir ? | 14:59.27 |
kens | No, I don't believe so | 14:59.34 |
| Though this was some weeks ago.... | 14:59.42 |
ray_laptop | and are you able to load CMap files: /Identity-H /CMap findresource | 14:59.53 |
kens | Didn't try to use CMaps. | 15:00.01 |
| If you want custom CMaps then I believe you would have to have a CMap directory inside the MyCIDFonts direcotry as well as a CIDFonts one | 15:00.29 |
| Or specify another parent to -I and put a sub-directory called CMap in there | 15:00.50 |
ray_laptop | kens: most people will want to ADD CIDFonts without having to replicate all of the CMap files (those they prefer to get from the %rom | 15:01.11 |
kens | Right, so you don't bother with CMaps. | 15:01.37 |
| You just do what I said above. | 15:01.42 |
| Make a folder, make a sub-folder called CIDFont, put the CIDFOnts in /parent/CIDFonts. Put cidfmap in /parent | 15:02.10 |
| specify -I/parent | 15:02.17 |
| Worked a treat for me | 15:02.24 |
ray_laptop | kens: I finished the comment. Please see bug 692300 | 15:02.25 |
kens | OK | 15:02.30 |
| Need to finish #692458 first, be a few minutes | 15:02.57 |
ray_laptop | kens: try gswin32c -I"%romesoufce/Init" -I/parent | 15:03.07 |
| kens: I suspect the only reason it is working is that you are getting the side effect that is the same as -sGenericResouceDir=/parent/ | 15:03.45 |
kens | But then I should get errors about a faulty GenericResourceDir. | 15:04.06 |
ray_laptop | but BOTH of those cause me to not be able to load CMap files | 15:04.09 |
| kens: I do get that warning | 15:04.20 |
kens | Let me finish this other bug first. | 15:04.21 |
| I didn't get that warning. But I can't concentrate on 2 topics, give me a minute | 15:04.37 |
ray_laptop | (unlless I have the -I"%romesoufce/Init" first -- then it doesn't find my CIDFont) | 15:04.52 |
| kens: np | 15:05.03 |
robin_watts_mac | ray_laptop: While you're waiting you could fix the topic :) | 15:05.27 |
ray_laptop | robin_watts_mac: actually, the 'fix' (1) does conform to the topic | 15:06.29 |
robin_watts_mac | Is the latest release not 9.04 now ? | 15:06.55 |
| (I meant IRC topic) | 15:07.01 |
ray_laptop | that is, to allow the cidfmap to have pathname strings (similar to Fontmap) to specify a location as with: /WadaGor-Bold (c:/Resource/myCIDFonts/WadaGo-Bold.cid) | 15:07.38 |
| robin_watts_mac: oh, that. I am having trouble regaining 'op' priveleges | 15:08.16 |
robin_watts_mac | Ah. | 15:08.43 |
henrys | can we please just remove the release when ... well when the permissions are fixed. | 15:08.45 |
| ? | 15:08.54 |
ray_laptop | henrys: sure, that's fine. | 15:08.57 |
kens | Hmm well they're right about the downsampling. I thought it was the threshold, but changin that had no effect. | 15:11.43 |
| OK I'll look at 692300 now | 15:11.56 |
| . | 15:11.57 |
henrys | ray_laptop:not very important but I was wondering if you saw my pagedevice question above - the cgm devices have no pagedevice dictionary and their is precedent in the code for not having the dictionary - just curious about it was going to ping Peter if you don't know. | 15:12.29 |
ray_laptop | henrys: ALL devices _should_ (must ?) have a pagedevice dict. | 15:13.13 |
henrys | one would think ... but see x_get_page_device for example | 15:14.16 |
ray_laptop | it's probably an OLD oversight. they should be using the 'SETPDEV2' macro (not SETDEV) and have 'page.dev' as a dependency | 15:14.43 |
robin_watts_mac | lolcat^^ and KindOne have arrived from #freenode to try and help us sort operator problems. | 15:14.57 |
lolcat^^ | We have? | 15:15.14 |
| I was hoping someone here was trolling... | 15:15.23 |
KindOne | eh... | 15:15.24 |
lolcat^^ | But anyways | 15:15.26 |
KindOne | tell your frind to login to nickserv | 15:15.33 |
lolcat^^ | Just do: /msg chanserv op #ghostscript lolcat | 15:15.49 |
| or whomever you want to op | 15:15.53 |
ray_laptop | henrys: if it is a 'vector' type device (not raster) then it could use the 'SETDEV2' macro | 15:16.12 |
| lolcat^^: I get: You are not authorized to perform this operation. | 15:17.06 |
| I remember doing something with chanserv before | 15:17.27 |
Erroneous | ray_laptop: Are you identified? | 15:17.44 |
ray_laptop | Erroneous: that ws it. Usually I get 'identified' by chatzilla automagically. | 15:18.37 |
Erroneous | ray_laptop: Now you can change the topic | 15:18.42 |
robin_watts_mac | Right. Now who else has op privs? | 15:18.44 |
| I seem to remember we had problems when ray tried to grant other people privs before. | 15:18.59 |
Erroneous | I wouldn't mind some... | 15:19.03 |
ray_laptop | Erroneous: how do we list those with op privs ? | 15:19.07 |
| I think henry is supposed to | 15:19.19 |
robin_watts_mac | In particular henrys and chrisl should be prived. | 15:19.24 |
ray_laptop | ahh the syntax to list flags is: /msg flags #ghostscript ray_laptop | 15:20.19 |
Erroneous | robin_watts_mac: /msg chanserv access list #ghostscript | 15:20.21 |
ray_laptop | chisl has +o | 15:20.42 |
henrys | if you want me to manage the topic I'll change it to something that never needs to be changed ;-) | 15:20.59 |
Erroneous | made the topic in #bitcoin-drugs | 15:21.21 |
robin_watts_mac | henrys and chrisl should have F and f too, I think. | 15:21.36 |
Erroneous | robin_watts_mac: why? | 15:21.57 |
robin_watts_mac | Because if Ray goes down with denge again, we're screwed :) | 15:22.15 |
ray_laptop | robin_watts_mac: as I said, chrisl has +o (op priveleges) | 15:22.44 |
| as does ghostgum. | 15:22.55 |
robin_watts_mac | (AIUI F and f are the 'top level' privs) | 15:23.04 |
ray_laptop | adding henrys now... | 15:23.07 |
henrys | ray_laptop:no there is something more to the setpagedevice stuff, we'd have the same problems if the device parameter IsPageDevice == false with X11 | 15:24.15 |
kens | ray_laptop I have done the following: | 15:24.53 |
| Created a folder in /Temp called cidfmap | 15:24.53 |
| The cidfmap file contains "/Test /MOESung-Regular" | 15:24.53 |
| I then start GS with gswin32 -I/Temp/cidfmap | 15:24.53 |
| I do not get nay warnings about GenericResourceDir. | 15:24.53 |
| I then do "(*) {==} 256 string /CMap resourceforall" | 15:24.54 |
| All CMaps are listed. Doign the same for /CIDFont lists MOESung-Regular and Test | 15:24.54 |
| executing "/MOESung-Regular-Identity-H findfont" works | 15:24.55 |
| Showing text doesn't, but then I don't know what woudn't result in /.notdefs for this font. | 15:24.55 |
| Drat, "/" caused problems | 15:25.17 |
| The folder "/Temp/cidfmap"contains a cidfmap file and a sub-folder CIDFont | 15:25.51 |
ray_laptop | kens: that's OK. Let me try it with different top names as you have (not c:/Resource) | 15:26.07 |
kens | OK | 15:26.15 |
| ray_laptop : you need a folder and a sub-folder called CIDFont. You specify the parent ot -I | 15:26.50 |
ray_laptop | kens: what do you have in your cidfmap ? | 15:27.02 |
kens | I toook it away and it still worked, so you can ignore that if you like | 15:27.21 |
ray_laptop | kens: OK. couldn't see why that would be needed | 15:27.46 |
kens | But originally I had one line: | 15:27.51 |
| "/Test /MOESung-Regular ;" | 15:27.52 |
| No, it doesn't seem to be required. | 15:28.00 |
| Butit does let you create substitutes for missing fonts, which is why I did it I thnk | 15:28.19 |
| to allow teh customer to define the Ordering+registry substitutes | 15:28.50 |
ray_laptop | kens: I get the GenericResourceDir warning when I do: gswin32c -Ic:/myResource | 15:29.08 |
kens | Hmm, I don't. | 15:29.17 |
| I'm not specifying c: though, let me try that. | 15:29.25 |
| Also not specifying a top level direcotry will try that too | 15:29.39 |
henrys | I'm inclined to remove cgm anyway given the comment in devs.mak and the popularity of the format. Objections? | 15:29.40 |
| anyone? | 15:29.47 |
ray_laptop | kens: I get the same warning with gswin32c -I/myResource | 15:30.06 |
kens | Weird, I don't. | 15:30.26 |
| FWIW chrisl didnt either IIRC | 15:32.01 |
marcosw_ | henrys: I have no objection to removing cgm. None of our customers use it, as far as I know. | 15:32.15 |
henrys | I guess 692456 has filled out their customer number is that right? | 15:33.36 |
| I hope not ... AIX compiler bugs are going to be time sink. | 15:34.00 |
| marcosw_ ^^^ | 15:34.23 |
ray_laptop | kens: something strange here. It seems that the directory MUST be called 'cidfmap', so that if I rename my c:/Resource to c:/cidfmap then (and only then) I don't get the warning | 15:34.26 |
kens | That is *indeed* odd! | 15:34.42 |
| so I struck lucky :-( | 15:34.54 |
marcosw_ | 692456 is indeed a customer bug, I've set it to P2. | 15:35.53 |
| does alexcher still have an AIX machine? | 15:37.37 |
henrys | marcosw_:well do you want to get a login and investigate. There is really nothing we can do with what they've given us. Well maybe run valgrind. | 15:38.14 |
marcosw_ | I'm downloading the eps file now and will try it under valgrind. | 15:39.01 |
henrys | wants to require a gnu toolchain on unix | 15:40.00 |
ray_laptop | kens: when I use -Ic:/cidfmap I don't get the warning, but it doesn't find my CIDFont | 15:40.05 |
henrys | but we don't yet. | 15:40.07 |
ray_laptop | henrys: won't fly with customers. Many have apps that are developed in their own toolchain | 15:40.45 |
kens | ray_laptop : your experience is at variance with mine, I can find and use the font. Can't see any text, but I don't know what CID to use. | 15:40.58 |
ray_laptop | and unix based cross compiles may not be gcc based (embedded systems often use other toolchains) | 15:41.22 |
| kens: trying again VERY carefully | 15:41.53 |
marcosw_ | the test file from bug 456 is an ad for a travel agent that offers trips to Vietnam. | 15:41.57 |
henrys | robin_watts_mac should fix it. | 15:42.28 |
| he knows what it should look like. | 15:42.56 |
| ray_laptop:I think all the embedding we've done recently has been gcc or windows. | 15:43.59 |
| robin_watts_mac would know | 15:44.18 |
alexcher | marcosw_: I have access to an old AIX box with gcc. | 15:44.27 |
marcosw_ | did it have some issue with not enough disk space or memory? | 15:45.08 |
alexcher | marcosw_: both. | 15:45.57 |
robin_watts_mac | Requiring gcc would be a retrograde step. | 15:46.46 |
alexcher | marcosw_: It's 32-bit, I had to use /tmp and switch off unused devices. | 15:46.51 |
robin_watts_mac | Lots of ARM users use RealView. | 15:47.07 |
marcosw_ | and I presume neither can be easily upgraded because they are IBM specific parts? Did you try eBay? I upgraded my old Sun Ultra 60 to 16 gigs of memory and added a second SCSI disk for something like $100. | 15:48.11 |
alexcher | marcosw_: It belongs to a 3rd party. | 15:48.44 |
henrys | alexcher:but you need the compiler to reproduce the problem right? or is it the cc compiler with the system? | 15:48.57 |
marcosw_ | never mind. | 15:49.03 |
alexcher | henrys: Of course, we need access to modern compilers and systems. The old box has gcc only. | 15:51.00 |
henrys | hmm a modern aix compiler, sounds peculiar. | 15:52.08 |
marcosw_ | I'll contact the customer to see if they'll give us access to one of their machines. | 15:53.18 |
henrys | I think if nothing shows up in valgrind we need a login and it's probably a waste of time building up a system to reproduce the problem, but whatever you guys want to do ... | 15:53.20 |
| robin_watts_mac:are there a lot of western tourists where you've been? | 15:55.00 |
robin_watts_mac | For the last couple of days, yes, definitely. | 15:55.20 |
henrys | I've always thought about going. | 15:55.28 |
alexcher | henrys: I will try to reproduce the problem on a PPC 32-bit first. | 15:55.39 |
robin_watts_mac | I've been in Cambodia at Angkor Wat and the surrounding area. Loads of westerners. | 15:55.46 |
henrys | alexcher:ah good point could be an endian issue. | 15:55.59 |
robin_watts_mac | Bits of Vietnam were full of westerners too. | 15:56.09 |
marcosw_ | henrys: valgrind didn't show anything, at least not writing to a ppmraw file (it looks like the command line from the bug is just viewing the file, I didn't try that yet). | 15:56.18 |
ray_laptop | alexcher: for AIX we have free access to systems at IBM | 15:56.30 |
robin_watts_mac | is currently in a bangkok airport hotel - about to go to bed so I can get some kip before the interminable flight home tomorrow. | 15:56.59 |
| But it's almost meeting time, so... | 15:57.11 |
henrys | it really isn't alexcher's bug he's just testing on ppc because he has one probably should be assigned to marcos | 15:57.52 |
| I really don't have much for the meeting today. I guess kens is still looking at alexcher's bug list kens? | 15:58.30 |
| ray_laptop:we looked at bug aging last week and thought alexcher's and your list were a bit long | 15:59.26 |
kens | doing Alex's lits, my own bugs, and this customer one from Ray | 15:59.27 |
henrys | ray_laptop:the memory stuff you took back ... I think to actually do that you are going to have to free *everything* - to continue past instance deletion right? | 16:01.26 |
marcosw_ | ray_laptop: yes, but I'm not sure if IBM still have AIX 5.3 available (I think it's pretty old) | 16:01.50 |
henrys | meeting topics from anybody? We can finish early if not. | 16:02.31 |
robin_watts_mac | would like to ask if anyone has a problem with us moving gs into line with lcms (w.r.t 16->8 bit conversion of colors) | 16:02.35 |
| The tricky bit is that gs has code for any 16 -> any depth, so I need to do some figuring to duplicate the effect. | 16:03.29 |
marcosw_ | running the file from the aix bug to the X11 device with valgrind over a slowish internet connection at Peet's is amusing. The text appears at 110 baud speeds :-) | 16:03.32 |
henrys | robin_watts_mac:I wanted to ask about that - what's up with doing line art conversion different than images? Is everything going through lcms? | 16:03.38 |
robin_watts_mac | Yes. | 16:03.46 |
marcosw_ | henrys: I filed a bug with Adobe for http://bugs.ghostscript.com/show_bug.cgi?id=690357 | 16:03.54 |
robin_watts_mac | For line art, it returns colors as 16 bit. | 16:03.58 |
| For images as 8. | 16:04.01 |
henrys | it == lcms ? | 16:04.19 |
robin_watts_mac | We then convert from 16 to 8 differently than lcms does, hence our resultant 8 bit value doesn't match lcms in all cases. | 16:04.38 |
| (Yes, sorry) | 16:04.42 |
marcosw_ | also, Adobe has been strangely quiet since I sent them kens' harshly worded reply to their last suggestion to 'refry' the pdf file to fix the text searching. | 16:04.44 |
kens | :-) | 16:05.02 |
henrys | robin_watts_mac:I am fine with making the conversion consistent - I guess a lot of changes though? | 16:05.40 |
ray_laptop | kens: the key to getting the -I switch to work without re-assigning the GenericResourceDir is that the path cannot contain 'Resource' anywhere, so c:/myResource is a problem, but c:/Rxsource is OK. If I have c:/Resourcex and use -Ic:/Resourcex then is _still_ sets the GenericResourceDir to c:/Resource/ | 16:05.41 |
marcosw_ | 110 baud might be optimistic. It's so slow that I can see each character being painted from top to bottom. | 16:05.55 |
kens | Oh, well that explains a lot ray | 16:05.58 |
robin_watts_mac | marcosw_: Register: WhatIsAnAdobeSupportControlWorth.com and post the entire email thread on there. | 16:06.05 |
| henrys: yes, potentially. | 16:06.11 |
mvrhel2 | Sorry I am a bit late | 16:06.17 |
robin_watts_mac | hi mvrhel2. | 16:06.27 |
mvrhel2 | hi robin_watts_mac | 16:06.36 |
henrys | marcosw_:Adobe bug filing has been about what I'd expect good luck. | 16:07.16 |
ray_laptop | kens: as long as it doesn't reassign the GenericResourceDir we seem to be OK. | 16:07.23 |
kens | Cool, will you tell the customer ? | 16:07.34 |
ray_laptop | kens: gswin32c -Ic:/Resource -sGenericResourceDir="%romesource/" also works (when the CIDFont is in the CIDFont subdirectory of c:/Resource) | 16:08.06 |
robin_watts_mac | will fiddle with rounding on the plane/in my jetlagged days when I get back then, at least to see if it solves the problem. | 16:08.24 |
ray_laptop | kens: gswin32c -Ic:/Resource -sGenericResourceDir="%rom%Resource/" also works (when the CIDFont is in the CIDFont subdirectory of c:/Resource) | 16:08.27 |
henrys | hi mvrhel2 we were just taling about lcms | 16:08.34 |
mvrhel2 | yes I see that. | 16:08.41 |
henrys | s/taling/talking | 16:08.42 |
ray_laptop | kens: (ignore the first one where chatzilla swallowed my % ) | 16:08.49 |
kens | yes, got that | 16:09.03 |
mvrhel2 | I think changing gs to do the rounding instead of the truncation is a good idea | 16:09.10 |
henrys | tor8:congrats on the new customer, anything for the meeting | 16:09.17 |
| ? | 16:09.18 |
| alexcher? | 16:09.22 |
alexcher | I don't have anything. | 16:09.51 |
mvrhel2 | robin_watts_mac: so I am working on fixing the pattern stuff so planar devices work | 16:09.56 |
robin_watts_mac | mvrhel2: I saw mention of problems on here. | 16:10.14 |
mvrhel2 | it is not just with the fast thresholding that it doesnt work | 16:10.16 |
henrys | alexcher:I did want to talk to you about using the gs allocator for jbig2 and jpeg2k | 16:10.26 |
robin_watts_mac | but haven't looked into it into any detail. | 16:10.29 |
henrys | but we can do that latter. | 16:10.31 |
robin_watts_mac | Right. | 16:10.31 |
mvrhel2 | hopefully I will have made some progress by the time you get back | 16:10.49 |
robin_watts_mac | mvrhel2: I should be home by this time tomorrow. | 16:11.08 |
mvrhel2 | for some reason I am having issues just doing my debug dump of the pattern bitmap though (which is planar) | 16:11.11 |
| the first plane I get ok | 16:11.28 |
henrys | mvrhel2:so regular planar rendering - before any of your changes didn't work with patterns? | 16:11.37 |
mvrhel2 | the rest are all zero | 16:11.39 |
alexcher | henrys: Let me commit a working version. The allocator cam be fixed later. | 16:11.42 |
mvrhel2 | that is correct henrys | 16:11.46 |
henrys | interesting but believable | 16:11.57 |
mvrhel2 | it did not work with the plank device and the old halfttoning | 16:12.00 |
robin_watts_mac | mvrhel2: It's unclear to me how much testing the old planar device had ever had. | 16:12.30 |
| (or rather the old planar memory code that my device wraps) | 16:12.47 |
henrys | but did it work with the old cmyk 1 bit and the device parameter? | 16:12.49 |
mvrhel2 | henrys: that I did not try | 16:13.15 |
| was that a real planar device? | 16:13.39 |
| or am I getting confused (which is likely) | 16:13.57 |
henrys | can't remember exactly have to look at the code. | 16:15.08 |
robin_watts_mac | For the ppm device, you could do -dUsePlanarBuffer and it would force it to use a planar buffer internally, and would convert to chunky for output. | 16:15.12 |
mvrhel2 | ok I will see what that does with my file | 16:15.26 |
| thanks | 16:15.29 |
robin_watts_mac | Maybe henrys is suggesting that the same could be done with the bitcmyk device. | 16:15.30 |
mvrhel2 | well, I belive we want to fix this issue | 16:15.52 |
henrys | right that's what I was thinking of the "plane extraction device" business. | 16:15.59 |
robin_watts_mac | My memory may be off here - grep for UsePlanarBuffer. | 16:16.01 |
henrys | I don't think it is important anyway we want planar earlier in the pipeline. | 16:16.26 |
mvrhel2 | yes | 16:17.09 |
| I agree | 16:17.11 |
robin_watts_mac | Unless anyone has anything else for me, I'm going to head to bed... | 16:17.18 |
henrys | have a good trip hom robin_watts_mac | 16:17.37 |
robin_watts_mac | Thanks. | 16:17.50 |
mvrhel2 | bye robin_watts_mac | 16:17.53 |
robin_watts_mac | I'll lurk as usual and read logs in the morning. | 16:18.00 |
kens | Night Robin | 16:18.07 |
henrys | okay I'm done but I'll be about if needed. | 16:18.44 |
mvrhel2 | ok. I am going to beat on the planar pattern stuff some more | 16:19.50 |
| although I need to look at that tiff bug | 16:20.05 |
| 692454 | 16:20.26 |
| that seems a bit surprising | 16:20.48 |
ray_laptop | kens: I closed the bug 692300 with the info. I'll let the customer that called me (780) know. | 16:21.15 |
kens | OK thanks ray | 16:21.26 |
| No, wait we still want paths in the cidfmap! | 16:22.02 |
marcosw_ | mvrhel2: yes, please fix the tiff bug, I use tiff output a lot for testing. | 16:23.05 |
ray_laptop | kens: not that I object, but why ? | 16:23.12 |
kens | because its nice to be able to specify the location of the CIDFont. | 16:23.31 |
ray_laptop | kens: OK, I'll re-open it as a P4 enhancement (which may never get done). I had started on modifying gs_cidcm.ps to detect stringtype and use that for the path. Probably not too hard to finish, but I have other more pressing issues. | 16:24.51 |
mvrhel2 | marcosw_: ok I will work on that now | 16:25.09 |
henrys | marcosw_:does old image alchemy do cgm? | 16:25.24 |
kens | ray chrisl had started on it laready | 16:25.26 |
marcosw_ | henrys: As i recall Alchemy can write cgm bitmap files, but not read them. | 16:25.48 |
marcosw_ | was wrong, AIX 5.3 is still available on the IBM Virtual Loaner Program. I'm requesting a loaner system now and will reassign the bug to me. | 16:26.26 |
ray_laptop | kens: by modifying the gs_cidcm.ps I hope ? | 16:28.34 |
| kens: the resource mapping scheme is really a mess (left over from Igor) but it isn't a big change | 16:29.37 |
kens | don't know ray. | 16:29.49 |
ray_laptop | I suspect that chrisl may have had to dig into it a bit for the FAPI stuff | 16:30.27 |
kens | very likely | 16:30.36 |
tor8 | henrys: the G customer I've been talking to signed on? | 16:33.12 |
henrys | well not yet but it seems promising | 16:34.05 |
marcosw_ | I have to run to school, anything else for me? | 16:35.37 |
henrys | not that I know of | 16:36.06 |
marcosw_ | okay, I'm leaving for Europe for 10 days on Thursday, I'll be in Spa, Belgium through the weekend and will likely have spotty internet access. Starting Monday I'll be in Germany and will be back online. | 16:38.27 |
kens | Have fun marcos | 16:38.37 |
henrys | okay so we should have a support fallback | 16:38.59 |
kens | you and me again ? | 16:39.11 |
henrys | that's fine for me. unless somebody else wants it. | 16:39.46 |
| kens:maybe chrisl and I can do it if you have stuff to focus on. | 16:42.26 |
kens | No more than usual ;-) | 16:42.44 |
| The only customer issues are Alex's | 16:43.29 |
| So its not a problem | 16:43.41 |
| Off for dinner now, night all. | 16:54.13 |
ray_laptop | kens: I sent the response off to the customer (780) and cc'ed you and chrisl | 16:54.16 |
| kens: g'nte | 16:54.21 |
kens | Right, thanks tay | 16:54.23 |
| ray | 16:54.24 |
mvrhel2 | crap. I didnt realize it was all tiff devices that don't work | 17:04.31 |
| oh it was the patch from Shailssh | 17:05.41 |
| Shailesh | 17:05.46 |
| at least it wasnt in the 9.04 release | 17:06.39 |
| whew | 17:06.41 |
| I wish I had caught this when testing his patch | 17:08.16 |
| Miles probably paid a bounty on that | 17:08.28 |
henrys | yup I'm sure it is already paid - which bug? | 17:12.11 |
| maybe the bounty policy should be the bug has to make it through the weekly test also. | 17:12.39 |
| is that what caught the problem? | 17:12.58 |
ray_laptop | I was on vacation, but that patch looked a little suspicious to me. | 17:13.52 |
| I had tried to add that comment, but then lost internet. | 17:14.27 |
| then it slipped my mind. ANY change in gdevdgbr.c without clear understanding is a worrisome | 17:15.05 |
| particularly since it was only for a wtsimdi device | 17:16.09 |
henrys | oh THAT change... | 17:18.52 |
mvrhel2 | http://bugs.ghostscript.com/show_bug.cgi?id=692300 | 17:25.00 |
| it passed through regression testing fine and it fixed the bug so it seemed ok to commit | 17:25.23 |
| so I did it.... | 17:25.28 |
henrys | that's about cid fonts | 17:25.50 |
mvrhel2 | oops | 17:25.57 |
| http://bugs.ghostscript.com/show_bug.cgi?id=687695 | 17:26.08 |
henrys | I see it the overprint stuff. | 17:26.14 |
mvrhel2 | yes | 17:26.17 |
| ray_laptop: that is why that one file was giving us a black image with tiff | 17:26.42 |
henrys | feel free to revert it and assign it back to him. | 17:26.45 |
mvrhel2 | ok I will do that | 17:26.57 |
| ray_laptop: it did not dawn on me that all of tiff was broken | 17:27.12 |
| ok. tiff devices should be working again | 17:41.05 |
| next time \me will commit what seems like an ok patch and let it fester for a couple weeks before closing the bug and saying it is ok | 17:43.02 |
| hmm had my control sequence wrong | 17:43.18 |
| back to patterns... | 17:43.52 |
cryptopsy | can mupdf automatically resize the doc so that it's dimensions are the same as the size of the mupdf viewspace? | 19:49.36 |
mvrhel2 | henrys: is mem_planar_copy_color_4to1 doing a 4 bit chunky to 4 bit planar copy? | 21:32.27 |
| it appears so. this is one of the issues with the pattern as it should be doing a copy color that is planar to planar | 21:34.01 |
| should be using the copy plane operation.... | 21:36.31 |
| but it has to go through the clip device | 21:37.19 |
| I think we need a few things in the tile fill state | 21:43.40 |
henrys | ah I'm back now. | 21:46.05 |
mvrhel2 | henrys: no problem. I think I am figuring this out | 21:47.23 |
| basically we are creating the pattern bit map ok as a planar object now but when we tile we are using the chunky to planar procedures | 21:47.47 |
| need to detect this and do copy_plane like we do with the halftone thresholds | 21:48.34 |
| I am hopeful this will be pretty straight forward | 21:49.01 |
| hmm tile clip device needs some work it looks like | 22:29.58 |
| I take retract my previous comment | 22:30.09 |
| I retract I mean | 22:31.00 |
| I was going to say take back but then changed it to retract | 22:31.16 |
henrys | what if planar always went through as clist? | 22:32.01 |
mvrhel2 | henrys: That would work | 22:32.20 |
| but would be slow for small tiles | 22:32.28 |
| for now at least we would get proper rendering though | 22:33.33 |
henrys | it might be worthwhile to get it going for now | 22:33.35 |
mvrhel2 | yes | 22:33.37 |
henrys | right | 22:33.39 |
mvrhel2 | let me see if it works for sure | 22:33.45 |
| before I say it does :) | 22:34.18 |
| finally found where it is decided | 22:46.58 |
| henrys: ok that fixes it | 22:49.10 |
| let me add in that check that if the target is planar we will force the clist | 22:49.24 |
| then we can do timing checks | 22:49.34 |
| with the color hallftone threholding | 22:49.49 |
| thresholding | 22:49.58 |
| then I will continue to get the non-clist case working. may want to drag robin into this when he returns and is over his jet lag | 22:51.05 |
| ok just doing a clusterpush to make sure. this really should not have any affect | 23:28.49 |
henrys | right, what was the reason for not always using the clist then upon playback if the pattern is small render to a memory device and tile the space with that? | 23:31.28 |
| the code would be much easier with that but I think there was some reason we can't do it. | 23:32.44 |
mvrhel2 | henrys: I do think playing back a clist drawing for a small little pattern was slower than simply tiling the bitmap. | 23:56.00 |
| Forward 1 day (to 2011/08/24)>>> | |