| | <<<Back 1 day (to 2011/12/14) | 2011/12/15 |
| kens | sebras you actually here ? | 09:17.08 |
| sebras | kens: yup, I am now. | 09:56.27 |
| | kens: why? | 09:56.29 |
| kens | Got an odd question from a free user, and tor8 and Robin aren't here ;-) | 09:56.48 |
| sebras | kens: though I'm at my job, so expect some response latency. :) | 09:56.49 |
| kens | No problem. | 09:56.54 |
| | Just a quickie | 09:56.58 |
| sebras | kens: sure. | 09:57.01 |
| kens | Do you know if MuPDF can read 'EmbeddedFiles' in a PDF file ? | 09:57.18 |
| | If the embedded file is a PDF ? | 09:57.26 |
| sebras | kens: no, it can't. | 09:57.31 |
| kens | Excellent, thanks. | 09:57.42 |
| sebras | kens: embedded files and file references are not parsed at all at the moment. | 09:57.49 |
| kens | Well, it's hard to see what it woudl do with them, with the sole exception of a PDF file. | 09:58.10 |
| | I do wonder what the point of embedding a PDF file in a PDF file is though.... | 09:58.28 |
| sebras | kens: possibly if it is an external file in the same directory, but I'm not sure about that. | 09:58.30 |
| | kens: yeah, only .pdfs would be considered anyway. | 09:58.49 |
| kens | :-) | 09:58.53 |
| sebras | kens: alright. have a meeting, got to go. | 09:59.06 |
| kens | THanks bye | 09:59.12 |
| sebras | kens: embedded pdfs seem to be rather uncommon, I have only seen pdfref do that (1.6 I think). | 10:34.25 |
| kens | Yes, my impression too, thanks | 10:35.47 |
| Robin_Watts | Morning tor8. | 12:33.18 |
| | I pushed 2 patches yesterday to my repo. | 12:33.24 |
| tor8 | hi Robin_Watts | 12:33.25 |
| Robin_Watts | I'm going to look at makinf fz_malloc scavenge from the store before it fails today. | 12:34.00 |
| tor8 | excellent! | 12:34.52 |
| sebras | tor8: good afternoon. ;) | 12:35.52 |
| tor8 | perhaps we should use size as a weighting metric for the eviction in addition to a pure LRU scheme? | 12:36.13 |
| | hi sebras. slow day at work? ;) | 12:40.13 |
| sebras | tor8: in between bugs. | 12:40.39 |
| | tor8: also -- how much should be evicted from the store when fz_malloc() is about to fail? 1.5 the size of the requested allocation? | 12:41.49 |
| Robin_Watts | sebras: The store has a max size, and a current size. | 12:53.45 |
| | If we need to allocate x bytes, and we can't, then I was planning to dump enough immediately so that max + x < store. | 12:54.49 |
| | If that isn't enough, then I'll dump max/16 from the store until it is. | 12:55.33 |
| tor8 | Robin_Watts: whitespace fixes on tor/context if you're in between commits. I can rebase it or remake it trivially later if you're already banging away on the scavenging. | 13:03.34 |
| Robin_Watts | Push yours to the master, and I'll pull from there. | 13:04.58 |
| | Well, I've written the scavenging code. Now I need to test it. | 14:44.23 |
| kens | Seems like the report file woudl be a good test. | 14:44.41 |
| Robin_Watts | The gre1970 - 01.pdf file was a nice way to test the store eviction. | 14:44.46 |
| kens | Ah... | 14:44.54 |
| Robin_Watts | But it's hard to run out of memory on an 8Gig computer. | 14:45.04 |
| | which swaps when it wants to. | 14:45.13 |
| | I might extend memento - have a MEMENTO_MAXMEMORY environment variable and let it fail all allocations over that amount. | 14:45.46 |
| henrys | Robin_Watts:what | 15:50.24 |
| | 's wrong with -K | 15:50.31 |
| Robin_Watts | mupdf. | 15:50.36 |
| henrys | excuse my linefeed | 15:50.45 |
| | oh okay | 15:50.48 |
| Robin_Watts | tor8: Do you think we need to keep the idea of 'ages' in the store any more? Or is the LRU and scavenging enough? | 16:35.47 |
| tor8 | no, I think we can drop the aging process | 16:36.21 |
| Robin_Watts | cool. | 16:36.32 |
| | Ok. Scavenging is working. I seem to be seeing some leaks from pdfdraw though, so I'll have a quick look for them. | 16:46.11 |
| marcosw_ | henrys and rayjj: I don't think we need a support meeting this week. I've looked at the new bugs and all are reasonably assigned. I'll be back online in an hour, so we can meet at 9:45 if I've missed something. | 16:49.19 |
| kens | Time to be off, goodnight everyone. | 17:11.54 |
| Robin_Watts | tor8: scavenging code pushed to my repo, for your comments. | 17:17.39 |
| | Still takes ~128Meg to render the big file though. | 17:28.51 |
| | but crucially, with 125Megs allocated, it makes no difference whether you render pages 1-10 or just page 9 - it fails due to lack of memory (and skips an image) in the same place. | 17:33.12 |
| | Reduces to 95Meg if I disable the use of the display list. | 17:39.08 |
| | so we could add an automatic fallback to non displaylist rendering in the app if we wanted. | 17:39.28 |
| tor8 | Robin_Watts: in the non-display-list rendering mode, we should only need to keep the image alive for as long as you need to render it. the display list will always gobble up lots of memory. | 17:41.05 |
| Robin_Watts | yeah. | 17:41.13 |
| tor8 | so we should probably make the display list non-default | 17:41.23 |
| Robin_Watts | The only way to get the memory use for this file down any further is going to be by adding partial image decode. | 17:42.08 |
| | And that's a big nasty. | 17:42.31 |
| tor8 | indeed. | 17:43.40 |
| | Robin_Watts: you probably shouldn't count substitute font data as part of the fontdesc->size | 17:46.55 |
| | since the substitute font data is static compiled in code | 17:47.50 |
| Robin_Watts | Won't the substitute font have a size of 0 then ? | 17:50.38 |
| tor8 | the other stuff (encodings and structs) that are part of the fontdesc count | 17:51.11 |
| | but the actual font data shouldn't | 17:51.17 |
| Robin_Watts | What line are you looking at ? | 17:51.44 |
| tor8 | pdf_load_substitute_font and pdf_load_substitute_cjk_font | 17:52.01 |
| | lines 201 and 220 | 17:52.21 |
| Robin_Watts | Gotcha | 17:52.35 |
| | Fix pushed. | 17:53.50 |
| tor8 | Memento_listBlocks ? | 17:54.53 |
| Robin_Watts | pants. Hold on. | 17:55.10 |
| tor8 | debugging printout that snuck in? | 17:55.11 |
| Robin_Watts | Ahem. | 17:55.40 |
| | I've forced a fix. | 17:55.45 |
| | yeah, sorry. | 17:55.51 |
| tor8 | looks good. I'll run it through sane and see if anything pops up. | 17:58.38 |
| | unless you already have? | 17:58.51 |
| Robin_Watts | Not yet, I need to transfer to the mac to do that. | 17:59.08 |
| | We should try memsqueezing the sane files to look for leaks. | 17:59.44 |
| | actually, let's try that now. | 18:04.21 |
| | laptop owners: You may want to consider installing http://preyproject.com/ | 18:05.02 |
| mvrhel | interesting | 18:06.32 |
| Robin_Watts | tor8: Just fixing warnings etc on the mac now. | 18:34.53 |
| | No pdfinfo build on windows... | 18:34.58 |
| | OK, unix build fixes pushed. | 18:40.11 |
| marcosw_ | Robin_Watts: besides being supported on non Apple hw how is prey different than find my iPhone? | 18:48.40 |
| Robin_Watts | marcosw_: Maybe it's not. | 18:49.01 |
| | Well, it's open source and free - dunno what "find my iphone" is. | 18:49.48 |
| mvrhel | how is prey free? it seems like they had a monthly charge | 18:51.34 |
| Robin_Watts | Prey Pro is chargable. | 18:52.46 |
| | Prey is free, as in beer, I believe. | 18:53.01 |
| marcosw_ | prey is only partially free, they charge up to $5/month if you want to have more than one device active (or use any of their other pro features, such as Active Mode which I believe find my iPhone does for free). | 18:53.14 |
| Robin_Watts | Just so you know: we will always offer free Control Panel accounts for anyone. | 18:55.14 |
| | They can hold up to three devices and store a maximum of 10 reports per device. | 18:55.16 |
| marcosw_ | that's odd, since one of the benefits of their $5/month plan is up to 3 devices. If that's available as part of the free tier it's not really a benefit. | 18:56.11 |
| Robin_Watts | That's not a benefit, it's a limit :) | 18:56.32 |
| | (see the bottom of that page) | 18:56.44 |
| marcosw_ | They say it's a benefit: "You can now upgrade to a Pro account and forget about the three device limit. " | 18:56.53 |
| | (sort of) | 18:56.58 |
| Robin_Watts | The personal account benefit is that you get 100 reports stored, rather than just 10. | 18:57.31 |
| henrys | marcosw_ : I sent you some mail about the customer that needs a response. | 19:37.54 |
| marcosw_ | henrys: will do | 19:38.38 |
| henrys | I say that to my wife when I don't hear what she said ;-) | 19:39.10 |
| Robin_Watts | henrys: I find that alternating "Yes dear", and "No dear" works better. | 19:39.52 |
| henrys | marcosw_:that PCL file is a real nuisance it is for ledger paper and there is no nice way of scaling it. My printer only goes up to legal. | 19:41.44 |
| marcosw_ | henrys: The internets (specifically wikipedia) says that the font that covered the wingings glyph ordering expired in 2005. If that's true is there a problem with shipping a wingdings compatible font? | 19:42.25 |
| | ^font^patent | 19:43.45 |
| henrys | marcosw_:I pinged URW a few years ago and they said they wouldn't do it. I didn't think it was 6 years ago though. I'll send them an email. | 19:44.19 |
| marcosw_ | thx | 19:44.47 |
| henrys | going out for my run - I'll check logs and email when I get back. | 19:45.56 |
| Robin_Watts | henrys: Stupid question here... can you not change the paper size in the PCL file and hope that the problem occurs in the visible region? Or does PCL allow you to position things relative to the other edges ? | 19:45.59 |
| | I guess I'm asking "Is PCL sane?", and I'm pretty sure I know the answer to that :( | 19:46.46 |
| henrys | sometimes this file does mostly absolute positioning so the font in question was always outside the clipping region after various changes to page size, orientation etc. | 19:48.51 |
| Robin_Watts | There's no way to do 'poster printing' with PJL ? | 19:49.15 |
| | (i.e. apply hard translations with the PJL that the PCL can't reset) | 19:49.41 |
| henrys | No there really should be a PJL command to scale to paper or do poster. | 19:49.56 |
| marcosw_ | henrys: what is the "new option to not embed fonts" you speak of? | 19:53.14 |
| henrys | marcosw_:kens put in something to use the platform fonts when emebedded fonts is false. | 19:54.17 |
| | best to test that on Windows | 19:54.53 |
| | bbiaw | 19:56.07 |
| penryu | in mupdf (v0.7), any way to advance from the bottom of one page directly to the TOP of the next page? | 20:53.31 |
| Robin_Watts | penryu: For a start, you might consider upgrading to 0.9. | 20:57.05 |
| | Many fixes etc in there. | 20:57.10 |
| | I can't promise that particular fix though. | 20:57.22 |
| penryu | assuming the Manual on the site is for 0.9, the feature isn't documented there. | 20:58.02 |
| Robin_Watts | To be honest, we are less concerned with the viewer itself, than the core libraries it uses. It's certainly possible to do that with the libraries. | 20:59.44 |
| | Writing a better GUI has been on our todo list for a long time (we've just done one for ios, for example) | 21:00.10 |
| | we haven't had time to do a better one for windows/linux yet. | 21:00.22 |
| kens | scratches head wondering what 'embedded font' feature henrys was referring to... | 21:00.36 |
| penryu | the lack of clutter around the viewing window is one of the best features of mupdf, imho | 21:00.51 |
| | Robin_Watts: thanks. | 21:05.35 |
| Robin_Watts | np. | 21:05.44 |
| | If you add such a feature, let us know :) | 21:05.53 |
| penryu | Robin_Watts: cloning now. maybe I will. | 21:07.59 |
| tor8 | penryu: we did tweak the key bindings a bit, I think there should now be a button to take you to the top of the next page. but as robin said, we don't spend a lot of time on the viewer. | 22:59.31 |
| penryu | tor8: ah, so I see. ,/. set panto = PAN_TO_BOTTOM/PAN_TO_TOP | 23:04.17 |
| | tor8: I guess I was thinking of overloading j/k to do that if there were no page left to pan | 23:05.11 |
| | tor8: thanks. | 23:06.07 |
| | Forward 1 day (to 2011/12/16)>>> | |
IRC Logs