| <<<Back 1 day (to 2013/03/14) | 2013/03/15 |
henrys | Mac Pro is down for the ssd installation | 01:06.03 |
| marcosw:the mac is up with the ssd now. | 02:31.10 |
marcosw | henrys: and is it lightning fast? | 02:31.21 |
henrys | isn't there something clustery you can do to find out? | 02:31.57 |
marcosw | probably. But that involves me doing something. Seemed easier to just ask :-) | 02:32.20 |
henrys | I'll be playing with it. | 02:32.49 |
| I guess I can run xbench and it will give me the performance of each drive | 02:33.36 |
| what is it you were using in Miami? | 02:33.48 |
marcosw | it appears to be generating the spotlight indices. mdworker and mds are both using 100% of a cpu. | 02:33.59 |
henrys | right I cloned the old drive so I expect that. | 02:34.21 |
marcosw | I used the blackmagic disk speed test (search black magic on the app store). It's free. | 02:34.32 |
henrys | will do, I'm going to wait until spotlight is done with its bullshit | 02:35.04 |
marcosw | I'm going to run a clusterpush, let's see how busy the cpus get. | 02:36.30 |
| htop doesn't seem to deal with your mac correctly. The even numbered CPUS never show any activity. And it shows all you ram in use (top shows 4 gigs free). | 02:38.38 |
henrys | where did you get htop from, /opt/local is woefully out of date | 02:39.57 |
marcosw | ~marcos/bin (I don't know where I got that one from, it's the one I run on my iMac) | 02:40.47 |
| it does now correctly show all cpus as busy, so I guess during the compile only 1 core from each cpu was being used. I don't know why, I use -j 12 so it should be using 12 cores. The mem value is correct as well. | 02:43.06 |
henrys | marcosw:well time for me to watch some tv. Modern Family is waiting, see you tomorrow. | 02:47.37 |
marcosw | night. | 02:47.49 |
marcosw1 | mvrhel_laptop: are you around? | 04:48.28 |
mvrhel_laptop | marcosw1: I am now | 05:41.27 |
| having fun trying to get a asynchronous tasks working in the windows viewer | 05:46.07 |
tkamppeter | chrisl, hi | 08:55.43 |
chrisl | tkamppeter: good morning | 08:56.32 |
tkamppeter | chrisl, I have uploaded GS 9.07 into Ubuntu Raring now, but with its onboard LCMS2. The Ubuntu maintainer of LCMS2 did not accept the changes from GS into the Ubuntu package. | 08:56.51 |
| chrisl, did you contact Marti about the problem, telling that what he has currently accepted makes GS segfault? | 08:57.30 |
chrisl | tkamppeter: mvrhel_laptop is the "staging post" for things going up to Marti. Having said that, I *think* the Fedora people as using code from Marti's repo with 9.07 - let me pull Marti's code, and have a look | 08:58.46 |
| tkamppeter: with the lcms code pulled straight from Marti's github repo, gs 9.07 works just fine with the file/command line you mentioned the other day. | 09:13.24 |
tkamppeter | chrisl, then out LCMS2 naintainer must have done something wrong. So for now I leave it the way it is and for 9.08 I will look whether the upstream lcms2 contains the new code. | 09:34.20 |
chrisl | tkamppeter: okay, but there is the worry that if the LCMS2 patches have been applied wrongly, or incompletely, other apps might have problems, too | 09:36.12 |
tkamppeter | chrisl, therefore we stepped back from applying the patches to our LCMS2 package and let GS 9.07 use the onboard LCMS2. | 09:37.51 |
kens | onboard in thsi case means the one we ship with GS ? | 09:38.17 |
tkamppeter | kens, yes. | 09:38.34 |
kens | Ah well that seems like a good solution to me | 09:38.44 |
chrisl | Yeh, I agree with that | 09:39.11 |
henrys | kens:support day for you. | 14:00.12 |
kens | Yes :-p( | 14:00.22 |
henrys | chirsl:we've got to figure out how to catch those before the release. | 14:00.59 |
chrisl | henrys: catch what? | 14:01.14 |
henrys | the ufst missing glyphs | 14:01.33 |
chrisl | that problem wasn't in the release | 14:01.44 |
henrys | oh than everything worked - yeah! | 14:02.05 |
chrisl | Well, not quite - the release code could get an overflow in the UFST rendering code - it was the fix for that that caused the very large glyphs to go missing. | 14:03.09 |
kens | I have to go for a dental checkup, back in a bit | 14:22.30 |
henrys | wow the pci ssd is almost 10x faster than another disk I have in the machine. Hope it helps the clusters. | 15:19.14 |
chrisl | henrys: holy cow! I might have spill for an ssd or two just for normal use! | 15:21.05 |
kens | I'm using an SSD for my boot device, its impressively quick | 15:21.27 |
henrys | the pci is much faster than sata but more expensive, marcosw led me down this road. | 15:24.03 |
chrisl | I'll stick to the sata interface for now. I might start with the laptop, though - better battery life | 15:25.18 |
henrys | I don't think they'll sell laptops without ssd's in a couple years. | 15:26.38 |
marcosw1 | henrys: the ssd has noticeably improved the cluster utilization on your macpro. I don't have hard numbers, but if I had to guess I'd say is 30% to 50% faster. Now I'm thinking of getting one :-) | 16:09.41 |
henrys | marcosw1:if I had to do it again though I'd get one size larger. 256G ain't what it used to be. | 16:14.49 |
mvrhel_laptop | hi kens: so it sounds like you found your issue which is an odd range scale | 16:22.45 |
| or different I should say | 16:22.52 |
kens | mvrhel_laptop : yes, but I don't understand why the ICC profile we create is wrong | 16:23.15 |
mvrhel_laptop | what happens when you use the icc_equivalent instead? | 16:24.20 |
kens | mvrhel_laptop : haven't got that working yet, should it work ? | 16:24.38 |
mvrhel_laptop | kens: If that does not work I def would like to know | 16:26.17 |
kens | OK well I will carry on with that. Of coruse that won't fic the bug in old versions | 16:26.44 |
mvrhel_laptop | no | 16:26.53 |
henrys | wonder why we don't just take the GPL code off the pdf995 site remove the nag screen build it for windows and release it for free on our site. | 17:08.36 |
mvrhel_laptop | sounds like a reasonable approach | 17:12.07 |
ray_laptop | who is the 'nathan' that ken "was asked to reply to" ? I don't see any email from him, and kens didn't attach the original email (Re: Ghostscript Support) | 17:25.17 |
| I am curious, because the response ken gave is (probably) incorrect and we may want to follow up | 17:25.55 |
kens | No it isn't incore | 17:26.09 |
| incorrect | 17:26.13 |
ray_laptop | kens: Oh, I didn't realize you were still here. Good | 17:26.36 |
| kens: doesn't pdfwrite still have the "automatic" selection of image compression based on a heuristic (data with "sharp" transitions selects Flate) ? | 17:27.55 |
| or did that get ripped out ? | 17:28.04 |
kens | That isn't eh point | 17:28.09 |
| He wants a JPEG in the input to be the *same* JPEG In the output | 17:28.23 |
| If we DCT the DCT data it will have new artefacts | 17:28.38 |
| Currently we have no 'pass through' | 17:28.54 |
ray_laptop | kens: That's not what I read, but I agree that is better, and we've had that request before. Probably an enhancement bug laying around for it | 17:29.33 |
kens | THat is what he was asking ofr Ray, as far as I can tell | 17:29.52 |
ray_laptop | kens: I read "preferably not modified" and DCT source to go into the PDF as DCT, and other image data to use Flate | 17:30.48 |
kens | ray_laptop : this isn't the way I read it | 17:31.00 |
| And we can';t do thaT either | 17:31.21 |
ray_laptop | kens: OK. BTW, does pdfwrite still do the automatic compression selection ? | 17:31.23 |
kens | Yes it does, but it has NO knowledge of the copmpression on input | 17:31.37 |
ray_laptop | kens: that's true. Pretty easy to add the compression type, however. Passing through JPEG unchanged is much harder | 17:32.40 |
kens | If we DCT the orifinally DCT image it will look bad | 17:33.01 |
ray_laptop | kens: I think that's what we do currently (in the default case) | 17:33.23 |
henrys | mvrhel_laptop: I saw a source code link on their page and thought they linked everything but they don't usually they just "exec" GS. | 17:34.14 |
ray_laptop | henrys: where do you see a source code link ? | 17:43.45 |
marcosw1 | alexcher: ping | 17:43.51 |
kens | goodnight all | 17:45.45 |
marcosw1 | henrys: just buy a second 240G SSD card and raid0 them, software raid on top of hardware assisted raid :-) | 17:46.05 |
henrys | ray_laptop:I confused the link for PSConvert.exe with their code. Bottom of the FAQ page | 17:46.42 |
ray_laptop | henrys: I see. That isn't their driver -- just the part that invokes gswin32c (AFAICT). The main part is still missing -- the source to their port monitor | 17:57.33 |
| which is what invokes PSconvert.exe | 17:57.56 |
alexcher | marcosw1: pong | 18:02.57 |
marcosw | alexcher: sorry, stepped away. I just needed you to install ImageMagik on alex_i7b. I sent you email as well. | 18:19.51 |
alexcher | marcosw: I cannot do it quickly. | 18:21.02 |
| marcosw: Imagemagic depends on some other packages that are no longer available. | 18:21.31 |
| marcosw: Perhaps, I need to upgrade the whole system. | 18:21.56 |
marcosw | alexcher: there isn't any hurry. | 18:31.38 |
| is alex_i7b the same operating system as alex_i7a? if so can you just copy the binaries and the libraries over? | 18:32.19 |
ray_laptop | chrisl_away: why did you need to change the name of the proc in the gx_io_device_procs_s ? (to gp_fopen) | 18:33.02 |
mvrhel_laptop | bbiaw | 19:20.42 |
ray_laptop | Yet another file that (probably) has the same funny way of painting -- fill with black, then paint an image over it (at least that's my bet with cust 531's latest problem report) | 19:23.35 |
| if the 'fill' is slightly too large, it paints over the image from the previous pair of operations, leaving the black. Depends on scaling and precise filling rules | 19:24.44 |
| and fill_adjust, of course | 19:24.55 |
| oops. I hate it when I make a simple change, then am in a hurry to get the cluster test started, but I forgot to test the build :-( | 19:30.34 |
| oh, great. x6 didn't get (or ignored) the abort :-( It's going to be really confused I think | 19:38.40 |
| yep. x6 is definitely confused. It is remaining idle | 19:42.03 |
| bbiaw. | 19:42.16 |
robin_watts_fone | waiting at Sheraton before check-in time. | 21:44.24 |
saper | which airport? | 21:45.16 |
robin_watts_fone | Miami | 21:48.35 |
| time to go all! | 22:17.35 |
| Forward 1 day (to 2013/03/16)>>> | |