| <<<Back 1 day (to 2013/08/05) | 2013/08/06 |
ray_laptop | I pushed my saved pages branch stuff to http://git.ghostscript.com/?p=user/ray/ghostpdl.git;a=summary | 01:41.18 |
| I emailed a request to tech to ask folks to look it over. This is for cust 801, primarily to satisfy their need for reverse order page printing (face up printing, they call it) | 01:42.16 |
| but I added some other halfway useful other stuff as well | 01:42.36 |
| please review and comment | 01:42.57 |
| I mentioned it to cust 532, and they are also interested in both reverse order printing and printing collated copies from the (relatively compact) clist | 01:44.23 |
| now they (532) spool compressed raster files to the disk, or re-parse and print the job | 01:45.06 |
| for reverse order, they go from the PDF (since they don't use our PS), but that is somewhat cumbersome to have to do it from a PS program (using runpdfbegin and dopdfpages) | 01:46.40 |
| whereas saved pages works with ANY iinput (PCL, PS, PDF or XPS) | 01:48.04 |
Robbes | good morning | 05:59.06 |
ray_laptop | Robbes: good morning. It's 11pm here on Pacific Daylight TImezone. Where are you ? | 06:18.23 |
ray_laptop | guesses Europe | 06:18.45 |
| but still quite early for the UK | 06:19.14 |
Robbes | yes europe, Netherlands | 06:25.28 |
| we have GMT+1 | 06:25.43 |
ray_laptop | Robbes: welcome to the gs and mupdf chat, then. | 06:46.32 |
| I haven't noticed you here before | 06:46.44 |
| Robbes: are you just browsing, are did you have something to chat about ? | 06:47.30 |
Robbes | rayjj: are you the same ray_laptop? | 10:52.25 |
| i've been playing around with gs on windows. | 10:52.37 |
kens2 | He is, and he's not here atm | 10:52.43 |
Robbes | I do indeed have one more question | 10:52.47 |
| ok | 10:52.53 |
kens2 | ask anyway, it'll get logged | 10:53.02 |
Robbes | hi kens2, thank you for you mail yerstarday | 10:53.03 |
kens2 | Ray is based in Los Angeles | 10:53.22 |
| So its pretty early in his morning | 10:53.36 |
Robbes | i was wondering,... if incomming PDF1.7 get's reformed into a new PDF 1.4, and the output can be conformed as PDF/A-2b, then.. didnt you loose features from PDF1.4+ ? | 10:54.00 |
| ah, i understand | 10:54.10 |
kens2 | thE pdf IS A BRAND-NEW pdf, AND THERE REALLY ISN'T MUCH IN TEH WAY OF MARKING OPERATIOSN THAT YOU CAN REPRESENT IN 1.7 THAT YOU CAN'T DO IN 1.4 | 10:54.36 |
| oops, sorry caps lock | 10:54.48 |
Robbes | ;) | 10:54.50 |
kens2 | there are extra compression types | 10:54.55 |
Robbes | ah, so what you are saying: 1.5, 1.6 1.7 only add some compression, no other functionals? | 10:55.21 |
kens2 | extra security options, additional interactive features, etc | 10:55.44 |
| But not really any changes to the graphics model | 10:55.53 |
Robbes | not like additional annotations, PAdES signature, transparrency, and maybe others? | 10:55.59 |
kens2 | Since pdfwrite only aims to reproduce the *appearance* we don't need to support those | 10:56.13 |
Robbes | ah, check. | 10:56.18 |
kens2 | transparency is in 1.4 | 10:56.18 |
Robbes | Oke, thank you. I know all i need to know :) | 10:56.45 |
kens2 | You're welcome | 10:56.51 |
Robbes | ah, but then it might be a good addition to alter documentation, for axample add that only PDF/A-1b en -2b are possible to create, and maybe add the b into the cmdline switch | 10:57.51 |
kens2 | I thikn we document that we only support 'b' conformance | 10:58.27 |
| And I'd prefer not to change the swicth as it breaks people usage | 10:58.53 |
Robbes | Ok. Thanks again and cu maybe later ~ | 10:59.37 |
kens2 | NP bb | 10:59.42 |
marcosw_ | henrys et.al.: I have a doctors appointment this morning and may miss the 9:00 meeting; if that happens I'll read the logs. The only thing I had for the meeting was asking tor7 about the seg faults in gxps from the new regression files that Robin_Watts checked in. In particular if I should disable the files pending fixes. | 14:42.49 |
tor7 | marcosw_: I have seen the reports, but not spent any time looking at the files or fixes yet. | 14:43.19 |
henrys | tor7:if you are not going to get to them soon make them bountiable shelly might look at them. | 14:46.44 |
tor7 | henrys: okay. | 14:47.05 |
| I'll take a quick look at them tomorrow then. | 14:47.28 |
henrys | meeting time | 15:00.51 |
paulgardiner | Right. Who's going first? | 15:01.53 |
henrys | mvrhel_laptop: windows apps? how far along are each? | 15:02.08 |
| I'm looking at the agenda | 15:02.21 |
| paulgardiner: did you have something in particular? | 15:03.11 |
paulgardiner | Yes - a problem I've run into with apearance streams for signature form fields, but no hurry | 15:03.55 |
henrys | tor7:hows opengl? | 15:06.16 |
tor7 | henrys: no real news to report. I mentioned I got the nvidia path rendering working, right? | 15:06.50 |
henrys | tor7:yes what is the next step? | 15:07.06 |
tor7 | my brain is melting in this heat, so progress is slow. | 15:07.20 |
| next up is text rendering, I think. | 15:07.37 |
| or a more general non-nvidia approach to path rendering, based on trapezoids | 15:08.00 |
henrys | hot where you are? | 15:08.02 |
tor7 | hot for where I am, not hot by US standards :) | 15:08.21 |
mvrhel_laptop | sorry I am late | 15:08.27 |
henrys | tor7:I just looked I don't turn AC on that for that. | 15:09.00 |
mvrhel_laptop | henrys: I had hit a snag in the windows phone last week, but figured out my issue yesterday. hoping to wrap up most of the windows phone features this week | 15:09.03 |
paulgardiner | Agenda?! I think I'm looking at the wrong thing. | 15:09.09 |
mvrhel_laptop | I want to get going on mupdf icc stuff after this | 15:09.28 |
henrys | paulgardiner: I think Robin created a link from the dashboard . | 15:09.43 |
tor7 | henrys: it's the threat of thunder hanging in the air. and high humidity. | 15:09.50 |
paulgardiner | Ah okay. That is what I was looking at. :-) | 15:10.16 |
henrys | paulgardiner: we did want to have signatures for Chicago so if you're stuck say something. | 15:12.01 |
| tor7:did you read about distance fields? worthwhile read for the rest of us? | 15:12.45 |
tor7 | henrys: it's an interesting algorithm that I want to try using. it's interesting enough on its own that it may be worth reading just for the sake of it. | 15:13.31 |
henrys | mvrhel_laptop: the phone was the first of the windows apps on the agenda right? not to be confused with windows mobile, which is the surface. | 15:13.58 |
| do I have that right? | 15:14.03 |
paulgardiner | I'm now able to create appearance streams matching those produced by Adobe Reader, but they include a logo, and it was only when I produced it myself (initially in black on white, rather than the usual light tint on white) that I noticed the logo includes a circle with and R in the middle. So I'm thinking we cant use it. | 15:14.42 |
henrys | tor7:I am also wondering about the svg reader? I wonder if zeniko would be interested in working on something a bit larger on a consulting basis? | 15:15.26 |
| paulgardiner: confused where is this logo coming from? | 15:16.56 |
paulgardiner | I took an existing appearance stream, as produced by Adobe Reader and edited it into mupdf path calls. | 15:17.47 |
henrys | all if you go to the regression dashboard and click on "work items" upper right under PCL it will go to the agenda. | 15:18.05 |
paulgardiner | The logo part that is | 15:18.06 |
mvrhel_laptop | henrys: we had the windows 8 app which I finished and is in the store, then the phone (mobile) app which I should have wrapped up shortly, then we have the desktop app which is going to be the replacement for gsview | 15:19.40 |
henrys | paulgardiner: I guess I don't understand why it wouldn't be easy to remove the logo | 15:19.49 |
paulgardiner | Yeah. We can just not create it, but then it doesn't look like a typical signed PDF | 15:20.30 |
| ... although Adobe Reader can be configured to not include the logo, so it would be acceptable. | 15:21.14 |
henrys | paulgardiner: probably being my broken record self but what does foxit do? | 15:21.23 |
paulgardiner | Good thought | 15:21.43 |
henrys | mvrhel_laptop: okay I did have 2 of them mixed up. | 15:21.55 |
| mvrhel_laptop: icc sort of lost steam with ubuntu using poppler - I'd lean toward the desktop app as being more important, what do you and others think? | 15:22.57 |
| everyone ^^^ | 15:23.04 |
mvrhel_laptop | henrys: yes, I agree | 15:23.50 |
| I think we need to get the desktop app done and I am in a windows api frame of mind these days | 15:24.13 |
henrys | mvrhel_laptop: sounds right to me. | 15:24.31 |
| mvrhel_laptop: I would like to get paulgardiner android stuff on the desktop app does it make sense for you you guys to work on a branch together or is it best for you to just pull and port from the android build? | 15:25.48 |
| I'm thinking for forms, signatures etc? | 15:26.01 |
mvrhel_laptop | henrys: it makes sense for me to just pull from what he has done. he is so far ahead in that area | 15:26.29 |
henrys | tor7:are you good with the priorities, do you see a pull for color management in mupdf? | 15:26.40 |
| mvrhel_laptop: okay | 15:26.48 |
| I guess that is good enough for the mupdf meeting, but I'll be around. | 15:28.27 |
paulgardiner | henrys: so far I'm failing to get foxit to sign the document. Are we sure it supports signing and not just checking? | 15:29.38 |
henrys | paulgardiner: hmm not sure. | 15:30.27 |
paulgardiner | It is displaying a pen icon in the filed, inviting me to do something, but I'm not sure what. | 15:30.55 |
henrys | from what I'm reading you need the sdk | 15:32.08 |
paulgardiner | henrys: I have to admit, it is now getting tight for Chicago. Possible if all that remains to do goes smoothly, but not if I hit a hiccup | 15:32.13 |
| henrys: I have to work out how to make openssl do pkcs7 signing, which I'm working on, and possibly close to doing. | 15:33.07 |
henrys | paulgardiner: like any other software schedule. | 15:33.15 |
paulgardiner | I guess. | 15:33.48 |
henrys | paulgardiner: don't worry about it, if it were urgent we'd put more people on it and slow it down more ;-) | 15:34.50 |
paulgardiner | On the logo issue, we could possibly have our own mupdf-specific one: Adobe have several quite different ones they use. Just to have some sort of logo makes it look more signaturey | 15:35.01 |
| henrys: Like it :-) | 15:35.14 |
| It's in that nice stage of feeling like it's winding up, but there is still about four tasks | 15:35.58 |
| Also, I should mention, I'm looking initially at just signing already existing forms with already existing unsigned signature fields, rather than support everything that can be done with signing. | 15:37.05 |
| ... not that the rest is necessarily a lot of work after that - just haven't gotten to thinking that all through | 15:37.47 |
henrys | paulgardiner: use ghosty | 15:39.59 |
| or the mupdf.com logo in the upper left. | 15:40.27 |
| I'm not particularly crazy about that but that's what we have... | 15:40.54 |
| it's better than the Artifex one which seems to suggest we can't draw a straight line ;-) | 15:42.04 |
paulgardiner | Yes, either sound good to me. I'd need to make them a lighter color so that black text can be clearly seen over them, and I'd need to get them in the form of mupdf path calls. | 15:43.08 |
| I guess I'll worry about it when the other tasks are complete, because no logo is also a possibility | 15:43.40 |
henrys | paulgardiner: right | 15:43.50 |
paulgardiner | henrys: is support for pdf signing in the windows app any use for Chicago? - because I'm likely to get there first before Android. | 15:51.20 |
| I mean the old windows app, not mvrhel_laptop's | 15:51.42 |
henrys | I think we are just showing mobile devices at the show | 15:52.20 |
paulgardiner | Okay | 15:52.49 |
mvrhel_laptop | is there anyway that I can get signing in the windows 8 app? | 15:54.34 |
| realize that I know nothing about this though, but my point is what makes it easy in the desktop app paul? | 15:55.21 |
chrisl_r61 | mvrhel_laptop: wasn't there concern over porting openssl to WinRT? | 15:56.07 |
paulgardiner | mvrhel_laptop: easier in the desktop app because the debugging is easier and I have yet to try building openssl for android. | 15:56.31 |
mvrhel_laptop | we would not do openssl but use what ever native api support there is in windows for digital signatures | 15:56.38 |
paulgardiner | mvrhel_laptop: as soon as I have something running on the windows app, I'll let you know in case there is an easy route to W8 | 15:57.05 |
mvrhel_laptop | paulgardiner: ok sounds good. | 15:57.25 |
paulgardiner | If you see any native apis for pkcs7 let me know. openssl is all I've been able to find. | 15:58.16 |
kens | Ghostscript meeting ? | 16:01.04 |
henrys | yes ray_laptop had a commit for us to review, maybe all of us can take some time today and have a look. I do think it worthwhile | 16:01.49 |
kens | I looked at it, but its not really my field, so Ididn't have a lot to say | 16:02.15 |
tor7 | henrys: (sorry for the delay) I'm a few days away from having ghostsvg ported to mupdf (in the same incomplete state as ghostsvg was) | 16:02.22 |
henrys | tor7:that's encouraging | 16:02.43 |
tor7 | henrys: I see no strong pull for color management in mupdf, so feel free to drop it back on the background priority projects | 16:02.49 |
| henrys: once I have that working, we could hand it off to zeniko or someone else. I think I'll still want to do the work on the text layout bits, seeing as they should be useful (in the long run) for the annotation creation that paul is doing | 16:03.38 |
henrys | I'm quite baffled by what customer 801 sent I can't figure out how it would work without a lot of changes. Has anyone else set up their system maybe windows is better. | 16:04.11 |
| tor7:okay | 16:04.21 |
tor7 | henrys: sebras has been doing some work on getting inline form field editing working on the x11 viewer | 16:04.28 |
ray_laptop | henrys: I haven't -- I was leaving it to you, but I can try it with windoze | 16:04.48 |
henrys | ray_laptop:no big deal i'll write to them today. | 16:05.13 |
kens | mvrhel_laptop : did you have a chance to look at that CIEBasedDEFG oddity that I sent you before I went on holiday ? | 16:05.23 |
henrys | I do have it running and all but I had to make a lot of changes. | 16:05.33 |
ray_laptop | henrys: I think you should send them the output you got with data on the Blue plane | 16:05.52 |
mvrhel_laptop | kens: sorry, I did not, ray dropped a p1 bug in my lap at the end of the week plus I was off 1 day. I will take a look at it this week | 16:06.01 |
kens | mvrhel_laptop : OK thanks | 16:06.10 |
sebras | tor7: I intend to upload my WIP changes to sebras/forms eventually. but not today. | 16:06.21 |
ray_laptop | mvrhel_laptop: and I am beating my head against the wall on the P1 bug. | 16:06.23 |
sebras | needs to go. | 16:06.24 |
mvrhel_laptop | ray_laptop: speaking of which, did you figure out the issue | 16:06.26 |
kens | I'd like ot get back to the colour work now I caught up on the bugs marcos dropped on me | 16:06.27 |
paulgardiner | tor7: I didn't realise he was doing that. Always assumed it would be difficult to do that in a way that matches the eventual appearance stream | 16:06.38 |
mvrhel_laptop | ray_laptop: were you able to duplicate what I was seeing? | 16:06.48 |
sebras | paulgardiner: what do you mean matches? | 16:07.08 |
| paulgardiner: I listen for keyboard presses and adjusts the text as desired and set the text of the widget. and do pdfapp_updatepage()...? | 16:07.37 |
henrys | kens:I was thinking the postscript/PDF group should take a quick look at the P1 bug assigned to me and waiting for Robin, just to make sure we understand the issue. | 16:07.40 |
paulgardiner | Looks identical... not that that is essential | 16:07.42 |
| sebras: oh right. Nice one. | 16:07.57 |
sebras | I haven't drawn the caret or handled mousing yet though. | 16:07.58 |
kens | henrys which bug number ? | 16:08.04 |
ray_laptop | mvrhel_laptop: I am seeing the clip for the coordinates and for the image position way above where the clip box is. Is an SMask supposed to paint with the previous clip in effect ? | 16:08.29 |
sebras | paulgardiner: see it as a prototype. once I'm done we know more of what makes sense and what does not. ;) | 16:08.29 |
henrys | http://bugs.ghostscript.com/show_bug.cgi?id=694451 | 16:08.32 |
kens | henrys I canlook at it I guess, I don't know that area of code well | 16:09.02 |
ray_laptop | henrys: that is cust 531's "squashed image with interpolation" issue ? | 16:09.08 |
mvrhel_laptop | ray_laptop: don't know. let me check the spec | 16:09.08 |
paulgardiner | sebras: I have some code as yet to be commited to golden that reworks my previous text layout efforts to work via the pdf device. Might be useful to you, possibly | 16:09.17 |
kens | ray_laptop : yes it is | 16:09.18 |
ray_laptop | henrys: is the image the correct size when you turn off interpolation ? | 16:09.22 |
kens | ray I'll try it | 16:09.44 |
henrys | ray_laptop: haven't looked. | 16:09.51 |
ray_laptop | kens: thanks | 16:09.51 |
kens | Oh, marcos hasn't attached the file, I'll dig it out of my m,ail archive | 16:10.07 |
henrys | chrisl: directory reorg? latest roadblock? | 16:10.38 |
chrisl_r61 | henrys: only that I hate nmake! It's progressing, bit by bit. | 16:11.10 |
kens | ray the customer syas that without -dDOINTERPOLATE it works correctly | 16:11.40 |
chrisl_r61 | henrys: I've got some issues with the rom file system to work out (for all builds) - at the moment the romfs always contains all files for all products | 16:12.00 |
ray_laptop | chrisl: I agree. nmake is a pain. The worst part is the way it differs in macro assignment (not lazy like "real" make) | 16:12.00 |
mvrhel_laptop | ray_laptop: I don't see any indication that the current clip would be ignored | 16:12.15 |
| in the spec | 16:12.22 |
chrisl_r61 | ray_laptop: yes, although it's mainly that I'm just not used to nmake | 16:12.36 |
marcosw | hello all, I'm back. | 16:12.45 |
chrisl_r61 | And, of course, I have to use Windows for it! | 16:12.52 |
ray_laptop | mvrhel_laptop: the thing won't work, then. But Adobe makes it work :-/ | 16:13.08 |
chrisl_r61 | marcosw: are we still on track to have the release branch test results on Friday? | 16:13.28 |
marcosw | chrisl_r61: yes, it's all going well. | 16:13.44 |
chrisl_r61 | Cheers | 16:13.56 |
ray_laptop | mvrhel_laptop: I am going to try zapping the clip to the BBox, instead of the clip in effect when we execute the SMask (when we do the fill with Gs7 in effect) | 16:14.20 |
marcosw | I'm on vacation for a week starting sunday, so the testing better be done by then. | 16:14.35 |
mvrhel_laptop | i was just thinking the same thing ray_laptop | 16:14.40 |
chrisl_r61 | marcosw: and you better not find any problems! | 16:14.51 |
mvrhel_laptop | ray_laptop page 234 seems to indicate the clipping path should be included | 16:15.32 |
henrys | URW would like to know if we are interested in Arabic and Hebrew additions to the font set. Have we had any requests? | 16:15.56 |
mvrhel_laptop | but that is the transparency group, not the softmask | 16:16.08 |
marcosw | henrys: I'm thinking of adding another cluster machine, to get us back to the 20 nodes we had with Alex's machines. Any thoughts? | 16:16.22 |
mvrhel_laptop | perhaps, since a soft mask is considered as a soft-clipping object, the clipping path is ignored | 16:16.39 |
chrisl_r61 | henrys: for the GPL or the 136 set? I'm not aware of any requests, though | 16:16.48 |
henrys | marcosw: miles' office | 16:16.50 |
| ? | 16:16.50 |
kens | ray I wonder if the problem is that we are outputting to 1 bpp, if I run to the display device with -dDOINTERPOLATE then the output is OK. The resolution seems to make no difference, 100 dpi has hte same result. I think the interpolation code doesn't expect to begoing to a 1bpp device. | 16:16.54 |
kens | is assuming tiffpack is 1bpp.... | 16:17.10 |
henrys | URW will not make anything GPL again I'm afraid. | 16:17.10 |
mvrhel_laptop | henrys: I have to head out a little early. have to run son to another day of computer camp. beautiful weather here, not the the best week for computer camp.... | 16:17.40 |
henrys | mvrhel_laptop: no problem. | 16:17.59 |
chrisl_r61 | henrys: I definitely haven't seen any customer requests for extra languages - I'm not even sure if any customers are using the 136 PS fonts | 16:18.09 |
marcosw | henrys: probably, though I may buy another dell poweredge server in which case it will have to live in my garage (they are too noisy for an office). The price has gone down and they are silly fast. | 16:18.22 |
henrys | marcosw: I was thinking of replacing henrysx6 with something more powerful would that be better than another machine? | 16:18.45 |
chrisl_r61 | henrys: But then, I haven't been around for *that* long....... customer(s) may have asked before I joined | 16:18.55 |
henrys | I don't know how antiquated henrysx6 has become. | 16:18.57 |
kens | ray_laptop : henrys the problem with the image is the device. Using the display device or tiff24nc there is no problem with the output. I assume that tiffpack is 1bpp and I'm guessing this must be the problem somehow | 16:19.27 |
henrys | chrisl_r61: I did have PCL requests ages ago ... | 16:19.34 |
chrisl_r61 | henrys: well, that's a different font set. I thought PCL was pretty clear on the character sets to support? | 16:20.36 |
ray_laptop | kens: hmm... interpolation for < 4 bits/per is handled differently | 16:21.05 |
kens | henrys ray_laptop the tiffg4 device behaves just like tiffpack | 16:21.12 |
| ray_laptop : I said I didn't know the code well :-) | 16:21.57 |
ray_laptop | marcosw: I still haven't bought one, but I think I will. | 16:22.01 |
henrys | ray_laptop, kens:anyway we've already told the customer it will wait for Robin I just though we should give it a quick look. | 16:22.08 |
kens | OK | 16:22.14 |
ray_laptop | kens: that is to be expected that g4 and pack (and probably pbmraw) all behave alike | 16:22.28 |
kens | ray_laptop : yes, just wanted confirmation | 16:22.38 |
| with a different device | 16:22.43 |
ray_laptop | kens: is it OK with -dNOINTERPOLATE to 1bpp device ? | 16:23.18 |
kens | ray_laptop : yes | 16:23.47 |
marcosw | henrys: there seems to have been a problem with moore's law. I replaced my 3 year old i7 with a new computer, a hex core i7 @ 3.4 GHz, it cost $1500 for the cpu/memory/motherboard/case/powersupply and is only slightly faster (my old i7 is running on the cluster as hubbles, the new one is called i7). The short answer is that it's probably not worth replacing henrysx6. | 16:23.48 |
henrys | 5 minutes left anybody want to raise stuff - the dashboard has a link called "Work Items" that points to the "todo" list. | 16:23.49 |
| marcosw: okay | 16:24.23 |
chrisl_r61 | henrys: I haven't heard anything back from Norbert | 16:24.30 |
marcosw | ray_laptop: the price from theserverstore on eBay is down to $409, that's for a dual Xeon system @ 2.8GHz | 16:24.32 |
kens | ray_laptop : I oculd look at the code, but I'm not sure I'd make any progress before Robin gets back anyway | 16:24.38 |
ray_laptop | marcosw: OK. That does sound pretty good. That's w/o a hard drive, right ? | 16:25.21 |
henrys | marcosw: how do the xeon system's compare with i7's? | 16:26.12 |
ray_laptop | kens: OK. I may have a look if I need a break from cust 532's bug | 16:26.29 |
kens | :-) | 16:26.35 |
marcosw | yes, but you use any >250 gig sata drive. I'm running my nodes with a cheap maxtor and a cheap seagate and both keep the cpu's 100% busy (probably wouldn't if we were writing bitmaps, but since the cluster jobs pipe to md5sum it's not using much disk bandwidth). | 16:27.17 |
ray_laptop | the powerdge has LOTS of ram, so that helps with tmp, right ? | 16:27.55 |
marcosw | ray_laptop: I just let Linux deal with caching, it seems to do a good job (I've tried running tmp in memory in the past and didn't see much improvement over having it on a drive). | 16:28.59 |
ray_laptop | I don't know how good linux is with using RAM to automatically buffer /tmp files | 16:29.17 |
marcosw | ray_laptop: like I said, the cpus are 95-100% busy when running a cluster job, so the answer is "good enough" | 16:30.15 |
ray_laptop | marcosw: I suspect it is pretty good, which is why the (sort of) slow hard drives don't help that much. Also the clist is designed to keep the disk from thrashing | 16:30.26 |
henrys | adjourned - I'll be around for a while thought | 16:30.54 |
| s/thought/though | 16:31.00 |
ray_laptop | henrys: OK. | 16:31.01 |
henrys | marcosw: where are the new macbook pros? it's been quite a while now. | 16:31.57 |
ray_laptop | reminder to all to have a look at my commit to http://git.ghostscript.com/?p=user/ray/ghostpdl.git;a=summary for the --saved-pages changes | 16:32.15 |
| (see the top of the logs) | 16:32.26 |
marcosw | henrys: they have 8 cores at 2.8GHz but with hyperthreading can run 16 threads. That's makes them nearly as fast as my 6 core i7 system which I have water cooled and overclocked to 4.3 GHz. At $409 (plus $75 for a drive) they can't be beat in terms of dollars/cluster job. If they only weren't so damn noisy. | 16:33.34 |
chrisl_r61 | ray_laptop: I take it that's a not quite ready to push commit? | 16:33.56 |
marcosw | henrys: I don't know when the macpro is going to be released, but it will be $3000+ for a reasonable configuration. | 16:34.16 |
ray_laptop | chrisl: well, I wanted some review and particularly comments on the questions I put in my email | 16:34.31 |
chrisl_r61 | ray_laptop: I just stumbled on the "#if 000000" in gsdparam.c, that's the only reason I ask | 16:35.12 |
ray_laptop | chrisl: I'd like to make sure that we all consider how we want it to work so we don't change it after pushing it, but before | 16:35.20 |
marcosw | the new macpro will be great machine for video editing/photoshop/etc but as a cluster node much of it's capabilities is wasted | 16:35.36 |
ray_laptop | chrisl: oops. Thanks. I'll fix that. | 16:35.37 |
marcosw | wonders what it would cost to rent rack space at a colo. he.net is 2 miles from my house... | 16:37.03 |
henrys | marcosw: the trashcan? | 16:37.05 |
marcosw | henrys: yeah, is that what we are calling it? :-) | 16:37.21 |
henrys | chrisl_r61: probably not unreasonable to get an ack from norbert. | 16:38.24 |
ray_laptop | chrisl: OK, I've fixed that to always IGNORE_INT_PARAM for PageCount, with the comment from the #if 00000 right above it | 16:38.31 |
chrisl_r61 | ray_laptop: thanks | 16:38.52 |
| henrys: I'll give it a day or so, and ping him | 16:39.08 |
marcosw | henrys: can I currently log into henrysx6? I have a spreadsheet of all of the cluster nodes that I maintain and might as well had your nodes (and peeves). If I can log in I can get the speed/memory/drive info/etc. | 16:39.21 |
ray_laptop | that's why we do reviews :-) I will also look over the diffs to see if I left anything else laying around in there | 16:39.36 |
henrys | marcosw: yeah it's 192.168..1.103 | 16:40.31 |
ray_laptop | marcosw: and you should be able to get into peeves | 16:40.51 |
chrisl_r61 | ray_laptop: I'll look over it again and read the documentation properly tomorrow morning - I've got a slight headache just now, so not concentrating too well..... | 16:41.29 |
ray_laptop | once I get a new server added, then I can take the plunge and upgrade the Ubuntu on peeves | 16:41.45 |
| chrisl: OK. Thanks. | 16:41.55 |
marcosw | henrys: thanks. | 16:42.48 |
kens | OK I'm off for tonight, goodnight all | 16:44.45 |
henrys | marcosw: I thought you could buy soundproof casing that was somewhat effective | 16:45.08 |
marcosw | uh oh, all of Miles' cluster nodes just fell over. | 16:45.35 |
chrisl_r61 | henrys: silent cabinets are *very* expensive, in my experience...... | 16:46.14 |
marcosw | except for points, which is on a UPS (along with the cable modem and router). | 16:46.19 |
| which is why bugs.ghostscript.com is still up :-) otoh, the ups only has about an hour of battery life, and we've been down for 25 minutes. | 16:47.14 |
henrys | chrisl_r61: silent yes but you can bring it does a few db's with something cheap I believe. | 16:49.45 |
chrisl_r61 | henrys: the problem with any kind of sound proofing/deadening is heat management - most sound insulation is also rather effective heat insulation. | 16:50.37 |
marcosw | everything is coming back up, odd I don't see any power failure messages from the UPS in the logs. | 16:51.16 |
chrisl_r61 | Power surge? | 16:52.58 |
henrys | bbiaw | 16:53.28 |
marcosw | chrisl_r61: perhaps, though I think there still should have been a log message. I guess I should read the documentation (I only tested the UPS by unplugging it, at which point it beeped and generated a bunch of log entries). | 16:54.26 |
ray_laptop | the best way to soundproof (IMHO) is to use "water" cooling for the case and CPU and then put the radiator somewhere else. The noise from modern disk drives is minimal | 16:55.42 |
chrisl_r61 | marcosw: I've had mains anomalies in the past (none recently - touch wood!) which would cause one or two machines to shutdown/reboot, but others carried on regardless..... | 16:56.06 |
ray_laptop | mvrhel_laptop: when you get back, please call. | 16:56.41 |
marcosw | forget to install the apcupsd software on the server (I installed it on the machine I tested the UPS on but not on the server in Miles' office). That explains the lack of log messages. | 16:59.21 |
| chrisl_r61: I think you may have been right, I've installed the UPS software and its internal log says the last issue was "Low line voltage" | 17:04.43 |
chrisl_r61 | marcosw: at least you know it's working well..... | 17:12.00 |
marcosw | is surprised at some of the colo deals, I can get 4U with $50/month. I would think the power cost is more than that... | 17:15.04 |
mvrhel_laptop | ok I am back | 17:20.36 |
ray_laptop | henrys: I just plowed through my email. Can you help on the questions from cust 801 (they all seem to be PCL related)? | 17:42.52 |
| henrys: I've got the P1 bug from 531 (that I just made a discovery on), then the P1 from 532 to look at. | 17:43.34 |
| mvrhel_laptop: As we discussed on the phone, I discovered that the cust 531 file works if I eliminate the clip in the parent group. Then the SMask is rendered as expected (unclipped). So I am going to set the default clip when rendering an SMask to the BBox, ignoring any outer clip, and restore the clipping when popping the mask. | 17:49.11 |
| errand. bbiaw | 17:49.35 |
mvrhel_laptop | ray_laptop: sounds good. let me know when you do a clusterpush | 17:49.38 |
| hi robin_watts_fone | 18:21.34 |
| off to lunch bbiab | 18:21.44 |
henrys | robin_watts_fone: no wifi? | 18:29.03 |
robin_watts_fone | hi. | 18:36.13 |
| rhino trekking tomorrow. | 18:36.30 |
| no laptop with me. in cabin. potential for lions between here and cabin. | 18:37.09 |
henrys | robin_watts_fone: sounds exciting, hope you see a lot of stuff. | 18:40.41 |
| bbiab | 18:43.11 |
robin_watts_fone | me too. just not lions too close :) | 18:43.28 |
| night all. | 18:50.56 |
marcosw | anyone have any thoughts on this generic error: | 19:02.09 |
| Processing pages 1 through 1. | 19:02.11 |
| Page 1 | 19:02.12 |
| **** ERROR **** | 19:02.12 |
| The PDF interpreter encountered an error, this file cannot be processed. | 19:02.12 |
| this is with gs master and going from pdf -> pxlcolor. It only happens on some of the cluster nodes. There is no other information (i.e. no usual PostScript stack). | 19:03.23 |
henrys | marcosw:is it repeatable on a particular node? | 19:40.56 |
mvrhel_laptop | ok raspberry pi just arrived | 19:42.10 |
henrys | marcosw:everyone should probably have ssh keys on each node and bugs can be reported against specific machines⦠then you could just report the bug and move on. | 19:42.43 |
ray_laptop | mvrhel_laptop: do you have an HDMI capable monitor, or are you going to hook it up to (one of) your home TV(s) | 19:50.54 |
mvrhel_laptop | ray_laptop: I have an hdmi monitor | 19:51.19 |
ray_laptop | cool. Mine is an OLD! monitor that only has VGA and DVI | 19:51.48 |
| maybe I should upgrade :-) | 19:52.08 |
| I took my laptop to the middle school for "career day", as I did last year, but since I upgraded laptop, this one only has HDMI out. The school is the last in the district being upgraded, so their projectors were all VGA only (which my old laptop had) | 19:54.01 |
| but I don't know if that is part of their upgrade or not. I hope so! | 19:54.42 |
| darn. Somehow, the image 'pcpath' is going back to the outer layer clip path, but I verified that the clip path was getting set correctly to the larger one of the SMask group BBox :-( | 19:59.16 |
| time for data breakpoints | 19:59.23 |
| oh. There is a 'setgstate' in '.execgroup' | 20:00.38 |
sebras | http://realbusinessatxerox.blogs.xerox.com/2013/08/06/always-listening-to-our-customers-clarification-on-scanning-issue/ | 20:01.11 |
mvrhel_laptop | ray_laptop: I have a vga and hdmi monitor hooked up here to the same machine. I had the same problem you had when scott and I visited a customer | 20:01.21 |
ray_laptop | so I actually have to modify the clippath in the GroupGState as well, I bet | 20:01.22 |
mvrhel_laptop | my laptop also has mini port output | 20:01.39 |
| and I got a mini port to vga convertor | 20:01.49 |
| which works well | 20:01.54 |
| I guess it is called mini display port | 20:02.30 |
| bbiaw | 20:02.45 |
ray_laptop | mvrhel_laptop: I see HDMI to VGA converters for $20-$40. I should probably get one | 20:03.17 |
| scott should probably get one for his iPad | 20:04.54 |
henrys | ray_laptop:yes I'll field the pcl stuff. | 20:05.49 |
ray_laptop | scott would probably like this one: http://store.apple.com/us/product/H9139VC/A/kanex-atv-pro-hdmi-to-vga-adapter-with-audio-support | 20:06.22 |
| wireless ! | 20:06.29 |
| henrys: thanks | 20:06.33 |
| If you want me to respond to any part of it, just say so in the response, and I'll pick it up | 20:07.07 |
teusje | can anyone spot a mistake in here http://pastebin.com/qYU6eex1 if I rotate 15° it rotates a lot more :s | 20:20.36 |
| https://skydrive.live.com/?mkt=nl-BE#cid=CB15F1A932B364BE&id=CB15F1A932B364BE%217339&v=3 | 20:20.43 |
marcosw | henrys: sorry, had to run out. | 20:28.52 |
| it's happens on peeves: | 20:28.57 |
| ~marcos/cluster/gs/bin/gs -o test.pxl -sDEVICE=pxlcolor -r300 -Z: -sDEFAULTPAPERSIZE=letter -dNOPAUSE -dBATCH -dClusterJob -dJOBSERVER ~marcos/cluster/tests_private/comparefiles/Bug689581.pdf | 20:29.00 |
| or simplified: | 20:30.04 |
| ~marcos/cluster/gs/bin/gs -o test.pxl -sDEVICE=pxlcolor ~marcos/cluster/tests_private/comparefiles/Bug689581.pdf | 20:30.09 |
henrys | if I were you I'd assign it to whoever fixed 689581, who could be better qualified ;-) | 20:30.43 |
| probably alex or somebody | 20:31.16 |
marcosw | henrys: yeah, I was mostly surprised (amused?) by the generic "**** ERROR ****" message and the lack of a PostScript stack trace. | 20:31.44 |
henrys | yeah that's kind of weird -Z# can help if you can filter through the noise | 20:33.09 |
marcosw | and by the fact that it doesn't fail during the nightly regression run. | 20:34.00 |
| ahh, there is the problem, it does fail during the nightly regression run, but doesn't return an error code so isn't flag as an error. I'm a bit surprised the cluster code detects this as an error, as I recall it uses the same code to read the log files as the nightly/weekly regression tests. | 20:37.44 |
henrys | just got back from this local place here, it's just like the soup nazi on seinfeld. The owner is deaf and pissed off all the time it's always an adventure going there. | 21:01.39 |
raven | Hi all! Tell me how to release memory in mupdf? If you render a many pages, gradually exhausted heap. How to fix it? | 23:45.54 |
marcosw_ | figured out why the cluster reported errors that the nightly regression didn't see. piping to md5sum is different than using lzop -d | md5sum. sorted, so starting tomorrow there will be new errors reported by the nightly regression tests. | 23:55.41 |
| Forward 1 day (to 2013/08/07)>>> | |