| | <<<Back 1 day (to 2011/12/21) | 2011/12/22 |
| Robin_Watts_ | malc_: I looked briefly at that, and wasn't hugely offended by it. | 00:13.04 |
| | I'll look again tomorrow. | 00:13.08 |
| | and talk it over with tor8 when I feel I understand it properly. | 00:13.54 |
| malc_ | Robin_Watts_: thanks | 00:28.11 |
| kens | Robin_Watts_ : I've cured the SEGV with language_switch, but its doing somethign stupid in deviceprops now and doesn't get past page 1. | 11:27.01 |
| Robin_Watts_ | kens: fun, fun, fun :) | 11:27.59 |
| kens | Make that .installpagedevice that goes pear-shped | 11:28.00 |
| | I'm not at all sure why, this works fine in the PostScript build | 11:28.28 |
| shin_ | Hi guys | 11:49.14 |
| | could give me some advice :D I added /Tele-GroteskEENor (D://te36013t.pfb) to my fontmap but on loading font i get sommething like that | 11:50.03 |
| | Loading Tele-GroteskEENor font from D:/te36013t.pfb... **** Warning: can't pr ocess font stream, loading font by the name. Loading Tele-GroteskEENor font from D:/te36013t.pfb... Error: /invalidfileaccess in /findfont | 11:50.11 |
| | ok -dNOSAFER my bad :D | 11:58.06 |
| kens | You in Eurpoe again marcosw_ ? | 12:39.54 |
| marcosw_ | yes, visiting the parents' for christmas. | 12:40.12 |
| kens | All your fmail y ? Nice. | 12:40.30 |
| marcosw_ | yes, we are all here, except for the guinea pigs :-) | 12:42.07 |
| Robin_Watts_ | tor8: ping ? | 12:57.27 |
| tor8 | Robin_Watts_: ping. | 12:58.26 |
| Robin_Watts_ | Bug 692714 | 12:58.36 |
| | I don't like his implementation particularly. | 12:58.48 |
| kens | Hmm, the language switch build doesn't allow us to override the built in resources with -I ? | 12:58.57 |
| Robin_Watts_ | I'd be tempted to extend the pdf_outline structure with a type enum and/or a flags word. | 12:59.40 |
| | We don't need to support all the different types (or in fact any more than we already do) immediately, but it would be nice to read the data out of the file for people who want to. | 13:00.47 |
| tor8 | Robin_Watts_: we should also get rid of the pdf_link type and replace it with a generic fz_link for use with xps | 13:00.51 |
| | muxps doesn't support in-page links yet | 13:01.00 |
| Robin_Watts_ | OK. I might look into that now. | 13:01.06 |
| tor8 | I've half started it when I added outlines | 13:01.21 |
| | so there is a list of link destinations, I just haven't extracted the actual links from the page contents though | 13:01.43 |
| Robin_Watts_ | Do you want to carry on with it ? | 13:01.44 |
| tor8 | I think both XPS and PDF link destinations could usefully target either a point or rect | 13:02.19 |
| | sumatrapdf hacked the new fz_outline to store a void* and a flag (is_pdf_link) so they can do their own extra bit of parsing | 13:02.56 |
| Robin_Watts_ | I think we'd be doing poorly to have whatever representation we pick being less expressive than PDFs. | 13:03.06 |
| tor8 | not too fond of that solution either, but I assume it's a stop gap measure while we figure out how to expose everything they need | 13:03.16 |
| | the PDF representation is all about taking away control from the user by forcing the viewer to behave a certain way | 13:04.02 |
| | the useful bit of data is the region of the page that should be visible, IMAO | 13:04.37 |
| Robin_Watts_ | We may not like the decision that PDF has made, but it's not our business to disagree. | 13:05.02 |
| | (or at least, the viewer can choose to disagree if it wants, but it shouldn't be the core lib that does so) | 13:05.24 |
| tor8 | xps only links to a region of the page, basically by targeting the xml nodes directly | 13:05.28 |
| | I'd be okay with enum + rect | 13:05.44 |
| | where the rect can be zero-sized for point locations | 13:05.58 |
| Robin_Watts_ | How about: point, point, zoom, flags | 13:06.07 |
| tor8 | zoom? | 13:06.20 |
| Robin_Watts_ | Where the flags have bits saying which of the points/zoom are valid | 13:06.24 |
| | In order to support /XYZ we need to be able to store point + zoom | 13:06.42 |
| tor8 | right | 13:07.35 |
| | for dumb clients, I think we could make a case that they should be able to always rely on say the first point then | 13:08.30 |
| | and if they want to get fancy, look at the flags and other values | 13:08.43 |
| Robin_Watts_ | tor8: the spec specifically says that in some cases null values mean 'don't change' | 13:09.07 |
| | And for a simple page link, there are no values at all. | 13:09.32 |
| | (we can set them to 0, but...) | 13:09.39 |
| tor8 | oh right, a dumb client would just use the page number | 13:09.40 |
| Robin_Watts_ | yeah. | 13:09.44 |
| tor8 | and a semi-dumb one would read 0 if we set it to that and it'd be the same effect | 13:10.03 |
| | and an annoying one would read the flags and do the right thing | 13:10.21 |
| | as long as they're set to sane defaults for clients that get it wrong I think we can go ahead with your scheme | 13:10.50 |
| Robin_Watts_ | cool. | 13:11.09 |
| tor8 | turn the pdf_link into fz_link and I'll take on the xps implementation after christmas | 13:11.40 |
| Robin_Watts_ | #ifndef _FITZ_H_ | 13:13.59 |
| | #error "fitz.h must be included before mupdf.h" | 13:14.01 |
| | #endif | 13:14.03 |
| | Why not, just #include "fitz.h" ? | 13:14.09 |
| kens | is lost in a maze of twisty passages, all alike :-( | 13:14.50 |
| Robin_Watts_ | Watch out for grues. | 13:15.18 |
| kens | Too late! | 13:15.36 |
| | Robin_Watts_ : you didn't start adding code to deal with unused parameters did you ? | 13:17.39 |
| Robin_Watts_ | kens: No. | 13:18.12 |
| kens | Thought not. | 13:18.16 |
| | It looks like htis code is trying to do something similar though | 13:18.28 |
| tor8 | Robin_Watts_: was following the plan9 header file organization structure where headers never include other headers directly. whether it's a good or bad idea, I don't know. at least it's an organization, unlike the situation with ghostscript's header files... | 13:18.56 |
| Robin_Watts_ | I believe that every header file should include every header file upon which it depends (and only those). | 13:20.31 |
| | Makes dependencies explicit. | 13:21.05 |
| sebras | Robin_Watts_: but so does #error... ;) | 13:21.23 |
| Robin_Watts_ | sebras: If I'm a user, and I know I want to use something in fitz.h, then I should be able to include fitz.h and rely on it getting everything else. | 13:22.12 |
| sebras | Robin_Watts_: though I do admit to siding with you when I discussed this with tor8 ~2 weeks ago. | 13:22.17 |
| | Robin_Watts_: so unless you type very fast you don't need to convince me. :) | 13:22.47 |
| Robin_Watts_ | But I'd also like there to be many small headers rather than 1 monolithic one :) | 13:23.09 |
| sebras | Robin_Watts_: what are the upsides of having many? from a programmer point of view, not a have-cpp-run-as-fast-as-possible point of view. | 13:24.14 |
| tor8 | there are a dozen ways to organize header files. I picked the lazy one. in a bigger project that took longer to recompile I would much rather have smaller header files (and correct makefile dependencies!!!!) | 13:24.16 |
| | broken build dependencies and small header files is much much much worse than anything else you can possibly do | 13:25.06 |
| Robin_Watts | ADSL reports 6.3Meg download. | 14:50.30 |
| kens | Sounds OK | 14:50.38 |
| Robin_Watts | If it stays like that, I'll be very pleased. | 14:50.44 |
| kens | Robin_Watts : commmit e391e19723bad3de19beb2fa584f0d9c8b62cac9 should finally fix the txtwrite device with language switch | 14:51.08 |
| | hough I still need to check PCL input | 14:51.16 |
| Robin_Watts | A BT engineer has been down the road replacing joints on all the circuits in the street all day. | 14:51.19 |
| | And a new BT engineer came here and replaced the joint outside my hour again. | 14:51.37 |
| | But I'm still suspicious that it'll all go to pot again within 24 hours, cos that seems to be the pattern. | 14:52.13 |
| kens | Not that there's any kind of fualt of coure.... | 14:52.14 |
| chrisl | Robin_Watts: how long has the ADSL been at your house? | 14:54.13 |
| Robin_Watts | I had isdn in 2003, and changed to ADSL 2 years or so after that. | 14:54.50 |
| | (as soon as it became available here) | 14:55.01 |
| kens | remembers ISDN | 14:55.05 |
| Robin_Watts | ISDN was twice as fast as modem, but it was "always on"! (Back when that was incredible) | 14:55.41 |
| kens | I know, I had it for some time | 14:55.56 |
| chrisl | Hmmm, a BT engineer told me (I wouldn't vouch for the accuracy!) that about 2-3 years ago there was a *huge* batch of faulty connectors for exterior use, that didn't seal properly. He said he'd been changing loads of them because the contacts had oxidised so badly. | 14:57.07 |
| kens | Hmm, anyone know how to drive the langguage switch build ? I want to feed it a PCL file | 15:03.08 |
| chrisl | You just give it the filename, don't you? I thought it worked out the type for itself | 15:03.43 |
| kens | I thought so too, but it gives me a -1 error | 15:03.56 |
| Robin_Watts | yeah, you should just give it the filename. | 15:04.23 |
| kens | Not a good sign then :-( | 15:04.32 |
| | Works if I don't use txtwrite :-( :-( | 15:05.19 |
| | Mind you, no text in that file either | 15:05.40 |
| kens | is confused now | 15:09.42 |
| | A different file (also without text) works fine. | 15:10.24 |
| Liquid-Silence | hiya kens | 15:13.05 |
| kens | Afternoon | 15:13.12 |
| | Hmm, font-FontType = 53 I don't think that's right ;-) | 15:13.54 |
| | Ug, its a PCL stick font.... | 15:14.30 |
| | No enumerate_glyph method :-( | 15:15.15 |
| mvrhel_laptop | good morning | 15:17.40 |
| Robin_Watts | What's the difference between /Fit and /FitB modes for Dest's in PDF files? (See page 582-3 of the 1.7 spec) | 15:18.09 |
| | Morning mvrhel_laptop | 15:18.14 |
| mvrhel_laptop | is marcos going to be around for the support meeting? I was going to ping him about doing the planar run with the null device | 15:19.53 |
| Robin_Watts | mvrhel_laptop: I think email is best. | 15:20.11 |
| mvrhel_laptop | yes probably true | 15:20.17 |
| kens | Robin_Watts : I suspect its historical, they are the same | 15:20.19 |
| Robin_Watts | He popped up very briefly on IRC, because he's in Europe, but I don't know if he's working. | 15:20.34 |
| mvrhel_laptop | oh. ok | 15:20.42 |
| | ok. now on to adding in the proofing profile | 15:28.43 |
| kens | Robin_Watts : looks like txtwrite needs some more work with PCL input. | 15:44.49 |
| | Hmm, pdfwrite doesn't seem to work too well with the language_switch build either. | 15:52.36 |
| | Exits with error code -28 | 15:52.48 |
| Robin_Watts | henrys did say that the language_switch build was not in a good stae. | 15:53.11 |
| | state. | 15:53.12 |
| kens | I think that may be an understatement | 15:53.25 |
| | I can see why teh txtwrite device is having a problem, so I thought I'd test the same file with pdfwrite to see how it copes with the problem. | 15:53.51 |
| | Doesn't seem to work at all with the language_switch and pdfwrite. | 15:54.06 |
| Robin_Watts | tor8: sanity check please? | 16:05.12 |
| | a pdf_link represents a place to go to within the current document, (or an action to be performed) | 16:05.43 |
| tor8 | Robin_watts: also external hyperlinks (http or another local pdf file, but we don't do the latter) | 16:06.56 |
| Robin_Watts | Oh, it's both ends of the link rolled into the same structure. | 16:07.08 |
| | i.e. it's both the source rectangle and the target. | 16:07.25 |
| tor8 | yeah. | 16:07.35 |
| Robin_Watts | OK, that makes sense. | 16:07.40 |
| | Morning marcosw_ | 16:22.30 |
| kens | Afternoon, he's in Europe! | 16:22.46 |
| marcosw_ | Robin_Watts: good evening Robin_Watts (I'm in Germany). | 16:22.47 |
| Robin_Watts | kens: Time is relative. | 16:22.55 |
| kens | Lunch time doubly so | 16:23.06 |
| Robin_Watts | This must be a thursday. I could never get the hang of thursdays. | 16:23.31 |
| kens | Why not have some tea ? | 16:24.26 |
| Robin_Watts | Just boiled leaves, in water? | 16:25.03 |
| | (I can play this all day) | 16:25.09 |
| kens | :-) | 16:25.12 |
| | What ? Where's the tea ? | 16:25.31 |
| marcosw__ | why does casper say "You don't exist, go away!" when I try to ssh to it? | 16:28.08 |
| kens | Existential crisis ? | 16:28.25 |
| Robin_Watts | marcosw_: Works for me. Are you trying to get in as the cluster or something ? | 16:28.40 |
| kens | Trying to log in from pinko commie European address | 16:29.25 |
| marcosw_ | Robin_Watts: I've tried both as me and as the regression use, both fail the same. I wonder if there is some issue with my parents' internet connection not having a valid reverse dns etnry or something, though it worked yesterday. | 16:29.49 |
| | odd, I can't ssh to any machine on the internets. | 16:30.10 |
| kens | See ? You *don't* exist.... :-) | 16:30.26 |
| marcosw_ | I'll renew my internet connection... | 16:30.31 |
| Robin_Watts | Can you get into your home machine and then out from there ? | 16:30.43 |
| shin_ | Hi :) today I have question like this :D | 16:30.59 |
| | if I have an embeded subset font in my pdf and I do ps2ps2 should I get that text worse in my postscript? Is it matter if I have this font maped in my fontmap gs? | 16:31.02 |
| kens | Define 'worse' | 16:31.21 |
| shin_ | hmm | 16:31.25 |
| | well when i print from pdf it looks ok, printed from ghostview is worse like dotted | 16:31.52 |
| kens | Fonts in fontmap will not get used at all for embedded fonts | 16:32.01 |
| | Sounds like it is halftoned or something. | 16:32.12 |
| | Could be somethign to do with teh text being coloured | 16:32.26 |
| henrys | well a foot of snow today it accumulates fast. | 16:32.28 |
| kens | What version of GS are you using ? | 16:32.35 |
| shin_ | 9.02 | 16:32.41 |
| kens | Well, 9.04 would be better | 16:32.55 |
| | But at this point, we need to see an example, so you should probably open a bug report | 16:33.15 |
| shin_ | you said sommething about color could that be the reason? this text is in footer and its gray like | 16:34.04 |
| kens | Text which is not 100% pure black will be halftoned. | 16:34.32 |
| | COUld be that you are getting a different halftone screen with the two versions | 16:34.46 |
| | How are you printing the PDF ? | 16:35.46 |
| shin_ | first of all I have a word then I do saveAs PDF -> then printing pdf from acrobat | 16:36.20 |
| | then ps2ps2 on pdf ant printing it from gsview | 16:36.31 |
| kens | Acrobat will probably be using the PPD to generate a halftone. ps2ps2 can't do that, os you get whatever default was present. | 16:36.59 |
| | Try changing the default halftone | 16:37.07 |
| shin_ | I will try to make this word entire black and see the result | 16:37.15 |
| | The black version looks the same for pdf and ps so as you said the problem is in color | 16:41.46 |
| kens | No, the problem is the halftone screen | 16:42.06 |
| henrys | Robin_Watts:There are likely other quality issues with language switch could the user specifiy the desired transformation and then we use the language or maybe it could just be pdf and ps input. | 16:42.11 |
| Robin_Watts | henrys: I could add an 'input language' set of radio buttons, sure. | 16:42.51 |
| shin_ | kens: well yes, so can I try to set sommething different to achive this good print in gray? | 16:43.21 |
| kens | it depends on whether the screen is defined in the PDF file or not. I would guess not. In which case you need to see if its defined in the PostScript fiel being output from ps2ps2 | 16:44.02 |
| Robin_Watts | shin_: I'd consider trying the latest version. | 16:44.27 |
| kens | If it is, then change the default in GS and you will get a different screen. If its not then you need to *set* a defult before sending the output to the printer | 16:44.32 |
| | Robin_Watts : don't think that will make any difference | 16:44.42 |
| Robin_Watts | There have been many color fixes in the latest version. | 16:44.49 |
| | DeviceGray rather than DeviceRGB etc? | 16:44.59 |
| kens | Robin_Watts : Its not colour, its the screen | 16:45.04 |
| | Or so I believe without seeing an example | 16:45.31 |
| shin_ | I'm suppries that on black version it looks ok even on the old pswrite method with -r300 | 16:49.48 |
| kens | pswrite is very crude. | 16:50.12 |
| | But black text is not halftoned, so it won't be a problem. | 16:50.46 |
| shin_ | well erier we were using some virtual printer to save ps from docs but now we have new requirements to save docx as pdf and the make ps from it....the problem is that with old gs pswrite methon on our one page pdf is huge like 2,5 MB with this -r300 its only 500KB | 16:52.45 |
| | earlier :D | 16:53.06 |
| kens | I told you, pswrite is very crude. | 16:53.07 |
| | All text is bitmapped for staerters | 16:53.15 |
| shin_ | ye you did | 16:53.17 |
| Robin_Watts | Is ps2write better? | 16:53.39 |
| kens | In what sense ? | 16:53.48 |
| Robin_Watts | Does it still birmap all text? | 16:53.59 |
| kens | Good god no. | 16:54.07 |
| | Text is preserved in fonts as vectors | 16:54.14 |
| Robin_Watts | Then shin_ might try using ps2write rather than pswrite? | 16:54.24 |
| shin_ | we're trying to move ahead with ps2write but printers in printing house cant print anything from ps2ps2 :/ | 16:54.30 |
| kens | He is complaiing baout the ps2write output, he's using ps2ps2 | 16:54.38 |
| Robin_Watts | oh, I see, sorry. | 16:54.48 |
| kens | He needs to get a proper halftone screen installed | 16:55.00 |
| Robin_Watts | or try the new version. (It's my drum, and I'm going to keep hitting it). | 16:55.27 |
| kens | Of course this is all rather guesswork on my part, with nothing to look at | 16:55.43 |
| shin_ | well I could upload that pdf and result from my ps2ps2 if you wold by that kind to take a look :D | 16:56.22 |
| kens | I'm off for the night shortly | 16:56.43 |
| henrys | shin_:with whom do you work? | 16:56.58 |
| kens | If you file a buig report I will look at it when I have time | 16:57.03 |
| Robin_Watts | shin_: You could go to bugs.ghostscript.com and open a bug and upload the file with details etc. But the first suggestion you get will be to try the latest version :) | 16:58.06 |
| shin_ | henrys: I'm just a programmer who is assign to help with transfering our client printing mechanizm to work with new requirements :D | 16:58.56 |
| Robin_Watts | shin_: But for what company? | 16:59.25 |
| shin_ | well I don't know really if I can tell who our client is :/ I can say I'm working in Comarch S.A. | 17:01.12 |
| kens | IT for Financial services ? | 17:01.57 |
| shin_ | not only generally it | 17:02.15 |
| | IT :D | 17:02.19 |
| | we have many sectors like telecommunication, financial etc | 17:03.12 |
| henrys | shin_:how is ghostscript integrated into this product you are supporting? | 17:03.13 |
| kens | Sounds like its being used 'as is' from a command line | 17:03.42 |
| shin_ | generally I don't know coz our client has the licence for it | 17:03.56 |
| henrys | should be working - distractions eat you alive. | 17:03.57 |
| kens | If your client has alicence, then they get support. | 17:04.18 |
| | So you should tell tehm to contact us direct, or give you permission to say who they are. | 17:04.35 |
| shin_ | I think that is resonable | 17:04.53 |
| kens | Well, you get better support that way ;-) | 17:05.15 |
| | We do support free users, but paying customers come first. | 17:05.33 |
| shin_ | :D | 17:05.53 |
| | anywa if they have this licence for old gs and they would decide that it time to go with 9.04 will they have to but new one? | 17:06.42 |
| kens | The licence doesn't work that way | 17:06.54 |
| shin_ | so how does it work? | 17:09.37 |
| Robin_Watts | If they have a license, then they get the latest version of gs. | 17:10.15 |
| | so it won't cost them anything extra. | 17:10.24 |
| kens | Depends if its a supprt licence or a redistribution licence | 17:10.31 |
| | But yes, both entitle them to the latest version and support. | 17:10.45 |
| shin_ | Ok I will try to get some details from them tomorrow | 17:11.19 |
| kens | Are you actually in Poland ? | 17:11.53 |
| shin_ | ye | 17:12.08 |
| kens | You are working late then ;-) | 17:12.22 |
| shin_ | well ye | 17:13.11 |
| | with polish salary we have to work twice longer then normal programmers hehe :D | 17:14.25 |
| | anyway did yuo have any reports saying postscripts from ps2write can't be printed on some printers? | 17:15.11 |
| kens | Well, time for me to go, goodnight everyone | 17:15.16 |
| | shin_ no. | 17:15.23 |
| | But many people use CUPS and that can inject PPD configurations into the ps2write output. | 17:15.55 |
| shin_ | hmm we have those report from printing house and I think they have some xerox or sommething like that | 17:16.20 |
| mvrhel_laptop | henrys: did you say you got a foot of snow? | 17:22.37 |
| henrys | yeah a bit more probably | 17:23.07 |
| | and still coming down. | 17:23.18 |
| mvrhel_laptop | wow. snowpack in the cascades is really low | 17:23.20 |
| | as in not much so far this year | 17:23.28 |
| | here in virginia it is 61 | 17:23.41 |
| henrys | that is unusually warm this time of year no? | 17:26.01 |
| mvrhel_laptop | yes | 17:26.05 |
| | I think the snow is finally supposed to start back up in the mountains when we get back home to washington | 17:26.35 |
| | hi ray_laptop | 17:59.18 |
| ray_laptop | hi, mvrhel_laptop | 17:59.44 |
| mvrhel_laptop | are you doing any travels over the holiday? | 17:59.57 |
| ray_laptop | mvrhel_laptop: no, my folks are out visiting, so we get to stay home :-) | 18:02.55 |
| mvrhel_laptop | thats nice | 18:03.02 |
| ray_laptop | cust 532's optimization effort is keeping me busy. They threw us a curve ball when they told us that the 600 dpi 2-bit (with the unoptimized multi-level threshold device) was their _default_ (from previous discussions, I thought it was a rarely used mode) | 18:05.16 |
| | then they keep throwing me these "fixes" that seem to fix the problem, but without understanding the underlying cause :-( | 18:06.43 |
| | and of course, the problems only show up on the target h/w | 18:07.27 |
| mvrhel_laptop | ugh | 18:11.29 |
| ray_laptop | taking a brief break from it now to work on the issue of closing and writing a PDF without exiting gs. It's looking like we don't need to have a .reopendevice operator, but can just change the gdevvec code to handle change of OutputFile differently (correctly) and %d format option | 18:11.33 |
| mvrhel_laptop | I did see the emails about the 2 bit stuff | 18:11.37 |
| Robin_Watts | tor8: Oh, ick. We have a ctx in every fz_obj ? | 18:11.38 |
| ray_laptop | mvrhel_laptop: yeah -- on one of the files, the 2-bit code is 12x to 14x slower to render. | 18:12.33 |
| | Hi, Robin_Watts | 18:12.38 |
| Robin_Watts | hi ray_laptop | 18:12.56 |
| ray_laptop | Robin_Watts: is the 'pattern' memento stuff described anywhere ? | 18:13.13 |
| Robin_Watts | ray_laptop: Not currently. | 18:13.25 |
| ray_laptop | Robin_Watts: so what is it ? | 18:13.34 |
| Robin_Watts | Normally, when we do memory squeezing, we get to a point and then fail all mallocs after that. | 18:13.57 |
| | and look to see what leaks. | 18:14.12 |
| | sometimes, we might have to fail several mallocs in order for it to fail. | 18:14.50 |
| | so the pattern stuff is an attempt to allow me to test more exhaustively. | 18:15.08 |
| | (i.e. test passing the first 200 mallocs, then failing 1, then passing 1, then failing the rest) | 18:15.40 |
| ray_laptop | Robin_Watts: a 'realistic' failure mode would be to fail subsequent mallocs that are some percentage of the failing one (smaller than the threshold would still succeed) | 18:16.27 |
| Robin_Watts | It explodes combinatorially if you're not careful. | 18:16.35 |
| | ray_laptop: Right, but memory squeezing is more about exhaustive testing than 'real world' cases. | 18:16.56 |
| | But I think using a PATTERN_MAX of 4 or 8 might still be useful. | 18:17.15 |
| ray_laptop | Robin_Watts: yes, I can see that trying "patterns" would expand the number of tries | 18:17.18 |
| Robin_Watts | But it's disabled for now. Maybe one day when gs and mupdf are perfect we can experiment with it some more :) | 18:17.59 |
| tor8 | Robin_Watts: only indirect refs I hope? | 18:20.05 |
| ray_laptop | Robin_Watts: but you are right that the current mode is more rigorous and can be 'real world' in a threaded environment where another thread has grabbed memory so even an allocation that was below a limit that succeeded previously may now fail | 18:20.14 |
| Robin_Watts | No, all obj's as far as I can tell. | 18:20.17 |
| ray_laptop | Robin_Watts: thanks for the explanation | 18:20.48 |
| | I'll let you and tor8 get back to work | 18:21.03 |
| Robin_Watts | ray_laptop: The idea is still (slightly) fluid in my head, so I haven't written it up in the memento source yet. | 18:21.13 |
| ray_laptop | Robin_Watts: I hope the fluid idea doesn't dribble out ;-) | 18:22.28 |
| Robin_Watts | ray_laptop: I'm holding my head level to try to avoid it. | 18:23.25 |
| ray_laptop | Robin_Watts: keeping a level head is always a good idea ;-) | 18:24.21 |
| Robin_Watts | henrys: Do you still run when it snows? (Not in 1 foot of it, obviously, but when it packs down?) | 19:22.05 |
| | I just bought some of these: http://www.snowandicegrips.co.uk/yaktrax-pro-ice-grips-2-c.asp | 19:23.19 |
| henrys | oh yes, I don't like yak trax I've an old pair of shoes that I put screws in which seems to work much better - at least for me. | 19:58.44 |
| | but for you that should work well - rocky trails will tear up yak trax pretty quickly. | 20:09.37 |
| | Forward 1 day (to 2011/12/23)>>> | |
IRC Logs