| <<<Back 1 day (to 2014/01/22) | 2014/01/23 |
marcosw | did someone reboot casper intentionally or did it crash? It's showing an uptime of 3 hours. | 00:44.35 |
ray_laptop | marcosw: I didn't reboot casper | 01:40.42 |
mvrhel_laptop | blah. finally I have this stupid managed code stuff calling into the c++/c unmanaged coded. not sure why it was such a pain. now I can hopefully make some real progress. I have been stalled in dll calling hell for the past 2 days | 07:10.05 |
| night all | 07:10.27 |
Robin_Watts | I'm going to be away from my computer for a lot of tomorrow. | 12:55.10 |
| I will be working from my parents in laws/a hotel over the weekend. | 12:55.34 |
| so please bear with any slow responses. | 12:55.59 |
kens | Hmm, have fun .... | 13:01.55 |
cg433n | Hello all. FYI the effect which I was trying to achieve yesterday with ghostscript is known as Gray Component Replacement. http://en.wikipedia.org/wiki/Grey_component_replacement | 13:33.12 |
Robin_Watts | ok. | 13:35.08 |
kens | OOh, finally got a decent description from the customer of what it is they are doing..... | 14:30.23 |
henrys | ugg a lot of shoveling today | 14:35.18 |
| kens: yes that was enlightening | 14:35.31 |
kens | persistence finally paid off | 14:35.56 |
| you have snow henry ? | 14:36.10 |
henrys | kens: yup not too bad 1/2 foot maybe | 14:36.59 |
kens | Better than Michael then. Still not sure how much we will have in Austria | 14:37.21 |
| Looks a bit bare in spots | 14:37.29 |
henrys | kens: the mountains are supposed to be good here. I live on the leeward side of the rockies we don't really get as much snow as folks expect. | 14:38.29 |
kens | 6 inches isn't bad.... | 14:38.45 |
| Wiw, Soll only has 45cm at the top of the mountain | 14:39.40 |
| And teh temperature is 0C | 14:39.59 |
henrys | we're supposed to have 12 tomorrow, shovel skipping is tempting | 14:41.27 |
kens | Henrys I'm meddling with gsprint at the moment, I think I cna add a media size parameter, and I *think* I can request that media from the printer, with a bit of digging I may be able to find out if the printer even has the requested media available. I'll spend another day or so poking it with a stick. | 14:41.54 |
| Finding out that the customer already knows the media size is a big win | 14:42.26 |
henrys | kens:okay how would you find that out/ | 14:42.47 |
| ? | 14:42.49 |
kens | By digging through the WINdows print documentation and trying out the API..... | 14:43.05 |
| THere's a 'DeviceCapabilites' call which is supposed to return this kind of information | 14:43.27 |
henrys | kens: cups of course has a database of ppd's but I imagine they are GPL | 14:43.41 |
kens | I think Windows can tell me itself | 14:44.01 |
| If i cna work out the right way to ask nicely | 14:44.41 |
henrys | kens: but I imagine each windows version is going to do it differently. | 14:48.33 |
kens | Henrys no its all the same API calls, at least from windows 95 to Vista | 14:49.06 |
| according to the SDK documentation | 14:49.18 |
henrys | kens:oh that's encouraging | 14:49.21 |
kens | Obviously this isn't going to work with WIndows 8 | 14:49.38 |
ray_laptop | kens: One thing I don't understand about Laura's description is that they say gsprint is faster sending a bitmap compared to gs mswinpr2. Since they are both sending bitmaps: | 15:05.54 |
| 1) can you confirm this, and | 15:05.56 |
| 2) If gsprint is driving the printer differently when sending a bitmap, is this something that can be ported to mswinpr2 ? | 15:05.57 |
| kens: BTW, good morning | 15:06.04 |
kens | ray_laptop : russel says this too | 15:06.08 |
| ray_laptop : I can't confirm it as I haven't tested it. | 15:06.33 |
| gsprint works by having GS render a bitmap to a pipe, which it picks up the end of and stuffs into a DIB on the pritner context. It then draws teh DC. | 15:07.12 |
ray_laptop | Russel said in the dcos that it is faster, yes. But I wonder how current that info is. That was probably written back in Windows 3.1 or 95 days | 15:07.17 |
kens | THisw is what I understand mswinpr2 to do also, so I'm baffled as to why there would be a performance difference | 15:07.31 |
ray_laptop | kens: AIUI, mswinpr2 also writes a DIB to a DC | 15:07.47 |
| oops. I typed too slowly | 15:08.00 |
kens | However, given we intend to dump gsprint, and presumably mswinpr2, in the futre, I'm not really keen o spending a lot of time on it | 15:08.13 |
| Now that I have a picture of what they do, and why, I'm focussing on getting gsprint to solve their problem | 15:08.42 |
ray_laptop | kens: true. | 15:08.52 |
kens | I would like to spend just a little time checking tghe assertion regarding speed. NB Laura said their tests also showed a performance difference | 15:09.13 |
| So its back to beting head against the API wall.... | 15:09.40 |
| I see fomr the logs Michael has been experiencing this amusement also. His comments remind me of when I wrote a Port Monitor :-) | 15:10.06 |
mvrhel_laptop | kens: fun stuff | 15:50.42 |
kens | :-) | 15:51.28 |
Robin_Watts | Gah. Too Many Hoops. | 15:58.05 |
| Now I've got to implement file numeration in my rom filing system. sheesh. | 15:58.20 |
mvrhel_laptop | bbiab | 16:20.55 |
ray_laptop | regression tests sure fly along with 28 machines, and particularly with the new fast ones. | 16:22.11 |
| finally getting really close on the pdf14 clist accumulation for other than gx_device_printer devices. Close enough that bmpcmp may show me useful info. I had forgotten that I had changed the BufferSpace default and that caused too many differences | 16:26.40 |
| so now I have reasonable results. AND NO SEGFAULTS ! | 16:27.12 |
Robin_Watts | nice! | 16:35.38 |
ray_laptop | well, I'll get to see how the 158 ps2write differences look at least | 16:39.16 |
chrisl | I think I'm going to stop writing release notes for Ghostscript - it's extremely clear no one actually reads them..... :-( | 16:57.30 |
kens | At least you can then tell people to go read them when they ask dumb questions.... | 16:58.01 |
| What prompted that ? | 16:58.05 |
chrisl | The mail about pdfopt.ps | 16:58.36 |
kens | Ah, haven't read it yet. Presumably they want to know why its missing ? | 16:58.50 |
chrisl | Exactly | 16:58.57 |
kens | I'd say just reply saying 'read the release notes' | 16:59.02 |
chrisl | I'll leave it to marcosw - he'd better at being polite than I am | 16:59.39 |
kens | Oh well, that is a known idiot | 16:59.50 |
| He could also try reading the ps2pdf.htm file which would even tell him how to do linearisation. | 17:00.39 |
Gigs- | need additional attachment marked private at http://bugs.ghostscript.com/show_bug.cgi?id=694874 thanks | 17:01.07 |
ray_laptop | kens: (or anybody else). Are any of the vector devices CMYK ? Or can pdfwrite or ps2write be set to present itself to the graphics library as CMYK and generate a CMYK (with -dCompatibiltyLevel=1.3) | 17:01.08 |
kens | ray_laptop : you can set the ProcessCOlorModel for pdfarie and friends | 17:01.26 |
Gigs- | chrisl: I may not read them as they come out but they are useful for figuring out when changes happened | 17:01.33 |
| I do use them | 17:01.38 |
ray_laptop | kens: does that work for ps2write ? | 17:02.01 |
chrisl | Gigs-: thank you, that makes me feel better... I wish there were more like you! | 17:02.14 |
kens | ray_laptop : with current code yes it shoud. You can also set -dColorConversionStrategy=CMYK to absolutely convert everythgin to CMYK | 17:02.33 |
| ray_laptop : In current code sttting ColorConversionStrategy to a device space will also set the ProcessColorModel | 17:03.16 |
| Or at least, it should...... | 17:03.31 |
ray_laptop | kens: great. Thanks. | 17:03.40 |
| I'll see what happens (checking in the debugger) | 17:03.57 |
kens | I don't seem to be able to set attachments to private any more | 17:04.00 |
chrisl | kens: are you logged-in? | 17:04.18 |
ray_laptop | kens: can you see private attachments ? | 17:04.25 |
kens | Hmm maybe not in that window | 17:04.29 |
ray_laptop | if you aren't logged in, you won't | 17:04.36 |
kens | OK yes I'mlogged in and I still cna't see how to make it private | 17:05.07 |
chrisl | Details->Edit Details and there's a checkbox in there | 17:06.08 |
kens | I don't seem to have an edit details | 17:06.32 |
| Actions:View | 17:06.44 |
| Duh found it | 17:07.07 |
| OK done | 17:07.14 |
| chrisl of course if the customer wants to use pdfopt.ps to generate a non-linearised file that merely claims to be linearised, they can still do so :-) | 17:09.01 |
chrisl | kens: that is true, but I'm assuming they'd prefer an actual linearised file | 17:09.33 |
kens | I can't think of any reason why anyone would want one..... | 17:09.46 |
chrisl | Because some daft customer insists it's a good idea? | 17:10.39 |
kens | Then the file that claims to be linearised but isn't is good enough :-0 | 17:10.57 |
chrisl | True | 17:11.11 |
kens | It may even get produced faster since we don't have to reprocess the whole file | 17:11.28 |
chrisl | Ah, "fast but wrong" - that's marketing gold, I tells ya! | 17:12.14 |
kens | :-) | 17:12.29 |
Robin_Watts | Better than slow but wrong. | 17:16.18 |
| unless it's a government contract. | 17:16.36 |
chrisl | I know lots of people who are often slow but wrong...... | 17:16.48 |
kens | OK goodnight folks | 17:31.49 |
ray_laptop | Well, down to 5 problem files with pdf14 clist accumulation | 17:41.19 |
| many expected differences due to rendering via clist, but that's OK. | 17:41.47 |
| oops. The point of this change is to avoid large memory allocations -- I forgot to disable that part :-/ | 17:53.48 |
Robin_Watts | paulgardiner: OK, romfss2 changes pushed to my epage repo. | 18:04.40 |
| Hopefully the commit message should explain everything. | 18:04.50 |
paulgardiner | Thanks. Perhaps it'll magically fix my current problems. | 18:12.44 |
Robin_Watts | ray_laptop: What sort of state is your office connection in? | 18:15.40 |
| Are you on a fast/unlimited connection now? | 18:15.55 |
ray_laptop | Robin_Watts: you mean peeves and peeved ? | 18:19.45 |
Robin_Watts | ray_laptop: Yes. | 18:19.54 |
ray_laptop | they are on a 7Mb downl 768K up | 18:20.22 |
| but it _is_ unlimite | 18:20.29 |
| d | 18:20.31 |
Robin_Watts | I can get the picsel test files from a friend. | 18:20.36 |
| But he's remote from me. | 18:20.46 |
| So either I need to buy an hd and get it shipped to him, and then have him copy the files and have it shipped back to me... | 18:21.10 |
ray_laptop | Robin_Watts: right, well, there's plenty of space on either machine, but peeved has a bit more iirc | 18:21.19 |
Robin_Watts | or I need to give him somewhere with 576Gig free that he can rsync to. | 18:21.28 |
ray_laptop | both are RAID-5 | 18:21.32 |
Robin_Watts | So, ideally, I'd want to set up an area on peeved that he could rsync to, and give him enough access to be able to do that. | 18:22.23 |
| and to ensure that it won't cripple your internet connectivity while it's running. | 18:22.35 |
ray_laptop | both have over 1Tb avail | 18:22.41 |
| turns out that peeves has just a bit more space (about 100G) | 18:23.16 |
Robin_Watts | My friend has a 10Mbps upstream connection, so he'd probably end up saturating your downlink, which would be bad for you. | 18:25.06 |
ray_laptop | Robin_Watts: either one is fine with me. Then I can put them on a USB drive to ship to marcos if he wants to put them somewhere where the uplink is better | 18:25.19 |
Robin_Watts | 576Gig USB drive? :) | 18:25.33 |
ray_laptop | Robin_Watts: sure -- I have a 2Tb USB drive (hard drive, of course) | 18:25.56 |
| _not_ a thumb drive unless your thumbs are really big | 18:26.34 |
Robin_Watts | I don't really know if we need these files, but I'm really loathe not to grab them while I can as otherwise we'll wish we had. | 18:26.35 |
| ray_laptop: Right. | 18:26.47 |
ray_laptop | Robin_Watts: right. You have root on them right ? | 18:26.58 |
Robin_Watts | I believe I do (on peeved at least). | 18:27.13 |
| If you're happy for me to play with that, I will go ahead and try and set up an area etc. | 18:27.29 |
ray_laptop | Robin_Watts: fine. Go ahead. | 18:27.55 |
| Robin_Watts: I confirmed that you are in sudoers on peeved | 18:28.24 |
Robin_Watts | ray_laptop: If I limit the transfer speed to say... 3Mbps would that leave you enough for day to day use ? | 18:29.18 |
ray_laptop | it may make peeves and peeved run more slowly on regression runs, but so what. | 18:29.39 |
| Robin_Watts: let's not limit it | 18:29.52 |
| your friend may want to limit his so that end doesn't die | 18:30.21 |
Robin_Watts | ray_laptop: Better to limit it a bit, I think, or you'll find things like bugzilla etc are unusable. | 18:30.33 |
ray_laptop | but I don't care about the transfer into them | 18:30.46 |
Robin_Watts | ray_laptop: His upstream is 10Mbps, so even if it saturated your end, he'd still have 3Mbps spare. | 18:31.06 |
ray_laptop | Robin_Watts: I am _rarely_ there. I am usually on my home cable modem (it's faster) | 18:31.14 |
Robin_Watts | ah, ok. | 18:31.24 |
ray_laptop | I think I stopped into the office last week to do something and check on the 5 pieces of junk mail pushed in the mail slot | 18:31.59 |
| so we are probably looking at 10 days or so of transfer. Are you sure it isn't worth buying him a USB 'backup' drive and sneaker netting it ? | 18:34.34 |
| or did I do the math wrong ? | 18:34.57 |
Robin_Watts | ray_laptop: That would get the disc to me. | 18:37.48 |
| It wouldn't get it to marcos. | 18:37.54 |
| and my connection is comparatively slow. | 18:38.08 |
| I'd have to carry it with me to texas. | 18:38.16 |
| which means we wouldn't get the data anywhere usable until the end of march. | 18:38.43 |
ray_laptop | Robin_Watts: OK. Go ahead with the transfer. I thought the disk could be shipped DHL or something. Particularly if you bought a new (couple of drives) and had the original packaging | 18:40.18 |
| Of course, you'd back it up before shipping one. | 18:40.34 |
Robin_Watts | ray_laptop: It cost me over 100 quid to ship a laptop to sweden recently. | 18:40.49 |
ray_laptop | drives are handy for backing stuff up anyway, | 18:40.59 |
Robin_Watts | international shipping is silly prices :( | 18:41.05 |
ray_laptop | Robin_Watts: WOW!!! really? | 18:41.10 |
Robin_Watts | yeah. | 18:41.15 |
| price of shipping would greatly exceed the price of the disc. | 18:41.38 |
ray_laptop | I guess so | 18:41.51 |
| I've shipped stuff to Venezuela and Mexico and it wasn't bad. | 18:43.07 |
| ahhh. That's better. A file that used 325Mb RAM (tests_private/comparefiles/Bug688728.pdf.ps.ppmraw.300.0) completed with only 14Mb now | 18:45.57 |
| for the ps2write step | 18:46.10 |
Robin_Watts | nice. | 18:47.20 |
| So, I just did a test and added some bates numbering to a PDF document | 19:21.55 |
| Headers/Footers are added to the document within marked content markers. | 19:22.41 |
| Those are BDC/EMC things that call a XObject form. | 19:24.27 |
| that XObject form contains the appearance for the numbers, and also a reference to some XML with the information in. | 19:25.03 |
| So, removing a bates number requires us to edit the stream. | 19:25.43 |
| henrys: My bit of the epage work is done for now (unless paulgardiner finds any problems with it), so I'm having a quick look at what's involved with bates numbering. | 19:30.48 |
henrys | okay I guess I should respond to Raed about that. | 19:31.44 |
Robin_Watts | I think extracting bates numbering can be done fairly easily; we'd need to look at a page, look at the form xobjects on that page, and search for one with the appropriate markers in it. | 19:34.25 |
| Adding bates numbering by adding a text annotation with the appropriate markers in is probably OK too (except for potential issues with what font we pick), but I suspect that will give us something that Acrobat can't 'remove bates number' on. | 19:36.03 |
| Adding bates numbers by adding to the contents stream itself is hard. If we can assume that it's a well formed stream, then probably we can wrap the existing contents in q ... Q and then put the required crap at the end. | 19:36.58 |
| but if it's not a well formed stream, we could end up making a mess. | 19:37.20 |
| Removing bates numbers from a document is again tricky. | 19:37.37 |
| We can probably textually run through the stream and find the blocks to remove, but again, it assumes that it's regularly formed; if the original document has the q/Q nesting wrong, then removing the block of text marked as calling the bates number will leave a broken file. | 19:39.03 |
ray_laptop | ALL: I'd like to get sp,e input on the user options to control the pdf14 rendering to using a clist accumulator (now that the basic functions mostly work). | 19:52.47 |
| So far I've moved the MaxBitmap from the gx_device_printer space_params to the gx_device (where MaxPatternBitmap resides), but the issue is that we also need to know the band size (unlike patterns). | 19:52.48 |
| Usually "BufferSpace" (also in gx_device_printer space_params) determines this. So, should I go ahead and move this as well ? | 19:52.50 |
| oops something got garbled on that first line. s/sp,e/some/ | 19:53.29 |
Robin_Watts | ISTR that something I had to look at a while ago would have been much easier if the space_params was in gx_device rather than gx_device_printer. | 19:54.39 |
ray_laptop | Robin_Watts: so moving the whole structure. That does help one ugly area I had of keeping the structure intact. I was having to copy the MaxBitmap into the structure so that the 'get_space_params' functions wouldn't change | 19:56.30 |
Robin_Watts | yeah, anything that makes stuff simpler has got to be good. | 19:57.35 |
sebras | btw, got the question today: why is the mupdf app called com.artifex.mupdfdemo..? | 20:09.56 |
| I don't remember the answer. | 20:10.13 |
| was it the case that someone else reserved com.artifex.mupdf..? | 20:10.33 |
Robin_Watts | sebras: Yes. | 20:12.30 |
| We used com.artifex.mupdf for the demo. | 20:12.40 |
| someone else built our demo into their app and uploaded it to the store. | 20:12.56 |
| hence they 'own' com.artifex.mupdf now. | 20:13.09 |
| We've had their demo removed, but we can't reuse that package name. | 20:13.27 |
ray_laptop | anybody need a 32Gb Surface tablet. Manufacturer refurbished. Only $169. | 20:45.04 |
| of course you can get an Asus Vivotab 32Gb (also runs Win RT) with the docking keyboard for $199 | 20:46.09 |
| but other than mvrhel_laptop I think we are all staying clear of Win RT units | 20:46.44 |
mvrhel_laptop | yes. I am on the bleeding edge as you say ray_laptop | 20:47.02 |
| and believe me I have cut myself a few times... | 20:47.13 |
| lunch time.. bbiab | 20:47.50 |
| ray_laptop: what day/time are your flights to texas? | 20:48.05 |
ray_laptop | mvrhel_laptop: haven't made my reservations yet. I guess I should do that. | 20:52.10 |
henrys | Robin_Watts: he seems to want the bates numbering in gs the way I read his email | 21:20.59 |
Robin_Watts | really? | 21:22.33 |
| I think he wants the ability to do all the 4 things I mentioned in my earlier message. | 21:23.42 |
| I told him it couldn't be done using MuPDF yet. | 21:23.50 |
| hence I think he's asking if it can be done using gs. | 21:23.58 |
| I think he just wants a solution and doesn't care where it's from. | 21:24.11 |
henrys | Robin_Watts: I guess we can talk to ken about it or just say no. | 21:37.09 |
| Robin_Watts: I assume pdfmarks and writing good pdf aren't coming about soon in mupdf | 21:37.51 |
| Forward 1 day (to 2014/01/24)>>> | |