| <<<Back 1 day (to 2012/12/04) | 2012/12/05 |
robin_watts_mac | mvrhel_laptop: See you in the lobby. | 02:16.11 |
chrisl | kens: ping | 08:42.06 |
kens | chrisl pong | 08:50.59 |
chrisl | That PCL font issue? So, it's a PXL file with a XL format embedded TTF font..... | 08:51.43 |
kens | Indeed. | 08:51.51 |
| The inquiry was about PCL-XL and PCL 5e so I explained they were stuffed on PCL-XL but 5e might work | 08:52.20 |
chrisl | The four other files in our regression suite I had marked as "to be investigated" for hinting issues are all also XL format TTFs | 08:52.33 |
kens | Aha, interesting | 08:52.46 |
chrisl | So, I'm thinking that some/many/all PXL drivers balls up the removing of the tables not needed by the XL format, and it breaks the hinting | 08:53.20 |
kens | It does sound like it doesn't it | 08:53.32 |
| And it makes it pretty clear that the HP printers aren't doing hinting (in some sense) | 08:53.50 |
chrisl | Well, HP ignore hints *at least* in such fonts - so I propose to do the same...... for now | 08:54.30 |
kens | Seems reasonable tome | 08:54.40 |
chrisl | I've already added code to do slightly different things for XL fonts in my not-yet-committed changes, so I'll make that change at the same time. | 08:55.56 |
kens | RIght, that'll be good. Will fix the problem for PCL rendering and also for pdfwrite, since it renxders such fonts. I suppose I ought to see why its doing that... | 08:56.34 |
chrisl | Without the required metrics tables, pdfwrite won't be able to build a valid TTF for the output | 08:57.20 |
kens | Yeah, but the conversion takes place before we know that (I htink) | 08:57.38 |
| I'll find some time to take a quick look today | 08:58.01 |
chrisl | I don't think it's that important - until you get time for the big TTF rewrite - then you'd have the option generating the missing tables yourself | 08:58.49 |
kens | yes, but goodness knows when that will happen.... | 08:59.05 |
chrisl | Well, as I say, there is no way (currently) you can embed the TTF from the file, so I wouldn't be too worried about it | 09:01.12 |
kens | I'm not concerned, just curious... | 09:01.25 |
chrisl | I have to head out - back in a few hours...... | 09:01.48 |
kens | OK bye | 09:01.54 |
| chrisl (for the logs) the reason that pdfwrite bitmaps the glyph is because the PCL routine pl_glyph_name returns -1. THIs is because there is no POST table in the TT font. Probably we could do better than that. | 10:17.56 |
| I ought to look and see what we do in the PostScript/PDF case when there is no POST table and port it over. | 10:18.28 |
robin_watts_mac | paulgardiner: Hey | 13:59.40 |
paulgardiner | hi | 14:00.03 |
kens | Hi robin_watts_mac | 14:00.15 |
robin_watts_mac | With reference to Paul Hudsons last mail, and chance you could extend the link stuff to do external links too? | 14:00.25 |
kens | Up early, jet lag or going for a run ? | 14:00.29 |
paulgardiner | Help! I'm not doing well helping out Paul hudson | 14:00.31 |
kens | disagrees | 14:00.39 |
robin_watts_mac | paulgardiner: I think you're doing fine! | 14:00.47 |
kens | Looks pretty good to me | 14:00.50 |
robin_watts_mac | IME hitting problems with customers, and then responding to them fast is absolutely fine. | 14:01.24 |
paulgardiner | robin_watts_mac: I'm happy to make that change too. There's some risk it will just leads to repeated requests | 14:01.31 |
kens | It always does, no matter what you do | 14:01.50 |
| Keep the code as examples for the next one to ask | 14:02.00 |
robin_watts_mac | paulgardiner: I was going to say, even if you just get it to the point where the link makes it into the java side, and he is left with the decision of "should I launch this or not", that would be massive progress. | 14:02.40 |
paulgardiner | Oh ok. I'll do that. | 14:02.40 |
| Yeah, I'd guess the JNI bit is what he doesn't fancy | 14:03.42 |
robin_watts_mac | If you get such code working, I'd be tempted to check it in, with the actual launch code commented out; maybe saying "Enabling this enables external links, but of course opens questions of security. Enable at you own risk." | 14:04.11 |
paulgardiner | Yeah. Makes sense | 14:04.33 |
robin_watts_mac | paulgardiner: Yeah. Given he's producing his own PDFs, I guess security isn't an issue for him. | 14:04.39 |
paulgardiner | I was unclear how much time I should spend on someone who isn't paying us, but while you are happy, I am. :-) | 14:07.00 |
kens | Scott and I think Miles seem keen that we should help them out | 14:07.25 |
robin_watts_mac | paulgardiner: He is paying us. | 14:07.25 |
paulgardiner | Ah right. | 14:07.33 |
robin_watts_mac | I thought... | 14:07.38 |
kens | Also, this is a good learning exercise for us, to find out what a real customer will want /need | 14:07.43 |
robin_watts_mac | maybe they aren't paying for support. | 14:07.46 |
| kens: indeed. | 14:07.54 |
paulgardiner | Come to think of it, I'm pretty sure external links were on a todo list somewhere. | 14:14.40 |
kens | Well, he's starting to sound pretty peeved... | 14:16.13 |
paulgardiner | Less so in the latest message | 14:16.47 |
kens | I gues that hasn't reached me yet | 14:17.00 |
| OK so he's just a 'fly off the handle' type. | 14:17.52 |
henrys | It was my view we should help out as much as possible. It is a new market that we are struggling to enter, but this guy is beginning to look like trouble and I'm questioning if we made the right move. | 14:22.18 |
robin_watts_mac | Swift change of plan. Biblical cloudburst. I'll run later :) | 14:23.35 |
kens | Hmm, what's the weather like apart from that, do I need a warm coat ? | 14:24.11 |
robin_watts_mac | kens: supposed to be 15-16ish. | 14:24.25 |
kens | OK cagoul and fleece will be fine then thanks | 14:24.38 |
| Hmm copying ghostpdl to my laptop WIndows says 'about 1 day remaining'.... | 14:25.22 |
paulgardiner | henrys: Good to have the confirmation thanks. Before, I'd been concerned that I might expend for time on it than you wished me to. | 14:25.58 |
henrys | paulgardiner:lets see how it goes. Good luck | 14:27.09 |
paulgardiner | :-) | 14:27.28 |
henrys | chrisl:so do we have a decision about hinting, you said it worked on the UFST but we don't scale this font with the UFST | 14:41.09 |
chrisl | henrys: sorry, phone call..... | 14:43.26 |
| henrys: So, all four of the other regression files which had hinting problems were also XL format fonts, so for the time being, I've made it so we disable hinting for XL format TTF fonts. Before the release, I'll try to work out a way to give a command line option to disable it for all fonts. | 14:45.08 |
henrys | sounds like a plan | 14:45.52 |
chrisl | henrys: oh, and UFST rendering of that font agrees with Freetype, *not* your HP nor the AFS code...... the hints are definitely broken! | 15:02.53 |
henrys | how are you testing that chrisl? | 15:04.02 |
chrisl | I can configure FAPI to pass TTF fonts to UFST | 15:04.32 |
henrys | okay because I think in the default UFST build now we are using free type for these downloaded fonts right? | 15:05.29 |
chrisl | Yes, if Freetype is built in, it will be used. | 15:06.31 |
henrys | okay I'm glad we got to the bottom of it ⦠in particular that it wasn't a bug in processing legal hints. | 15:08.14 |
| marcosw_:ping | 15:10.25 |
marcosw_ | henrys: monring | 15:10.31 |
henrys | what wireless router are you using these days, looking to get something that support 802.11n, I still have your buffalo device. | 15:11.19 |
chrisl | henrys: I just bought a Netgear WNDR3800 and put Open-WRT on it - and I'm pretty darned impressed so far...... | 15:12.35 |
henrys | Is it me or does everyone screw up rsync syntax and copy foo to foo/foo? | 15:12.48 |
| chrisl:thanks I'll look at that one. | 15:13.26 |
chrisl | henrys: note that I can't vouch for it with the Netgear firmware, I got it specifically to put Open-WRT on it | 15:14.27 |
marcosw_ | henrys: I'm using an Apple Time Capsule. It includes a hard drive for backup; also works really well with an Airport Export in extending my network to the other part of the house. | 15:14.56 |
| but I agree with chris, Open-WRT is the way to go. | 15:15.09 |
henrys | okay I've given apple enough of my money anyway | 15:15.41 |
marcosw_ | (or tomato, assuming it's still being developed). | 15:15.50 |
henrys | chrisl:does that have vpn support? | 15:16.25 |
robin_watts_mac | If people are looking at a time capsule solution without paying Apple, then I recommend the Netgear ReadyNAS NV+ v2 | 15:16.40 |
chrisl | henrys: Yes, but don't ask me how to set it up..... it's basically a mini-linux computer | 15:18.09 |
robin_watts_mac | It's a 4 slot NAS box that works with rsync/Samba/Time capsule etc. You can add new drives as you need more capacity, and as long as you always add drives at least as big as the ones you start with, it handles all the RAIDing for data safety for you. | 15:18.16 |
| (It's a NAS, not a wireless router though) | 15:18.35 |
marcosw_ | robin_watts_mac: welcome to sunny california :-( | 15:19.01 |
| where are you staying? | 15:19.10 |
robin_watts_mac | Hyatt. | 15:19.15 |
marcosw_ | near the airport? Is that the meeting hotel? | 15:19.24 |
robin_watts_mac | yes. | 15:19.28 |
henrys | I use hamachi to get to my computer remotely but I might look at doing vpn on the router. | 15:20.35 |
marcosw_ | I went the other way. I run an apple time capsule as my router and FreeNAS on my NAT (or is the ReadyNAS NV+ firmware open source?). | 15:21.04 |
| s/NAT/NAS/ | 15:21.17 |
henrys | robin_watts_mac: when is the meeting? | 15:21.38 |
robin_watts_mac | Meeting Miles outside at 11am. | 15:22.21 |
| meeting is 11:30 | 15:22.25 |
henrys | good luck hope you can stay alert after the long flight. | 15:23.19 |
robin_watts_mac | me too. Though if they ask awkward questions, I may just snore and leave them to Michael :) | 15:23.57 |
mvrhel_laptop | ha | 15:24.13 |
| I will give you a poke now and then | 15:24.26 |
henrys | mvrhel_laptop: bring your smelling salts | 15:25.17 |
mvrhel_laptop | :) | 15:25.29 |
marcosw_ | miles invited me to the meeting to sit on the other side of robin_watts_mac and poke him as well | 15:29.31 |
mvrhel_laptop | hmm it does look a bit like seattle weather outside | 15:29.52 |
| but a few degrees warmer | 15:30.29 |
henrys | marcosw:guess we'll have a lot of people ⦠if it comes down to a vote we should be good ;-) | 15:36.01 |
mvrhel_laptop | 1/2 the company will be there | 15:39.31 |
robin_watts_mac | suspects that marcosw was being flippant. | 15:39.56 |
mvrhel_laptop | I was wondering about that | 15:40.39 |
| robin_watts_mac: if you want to eat earlier I am fine with that | 15:40.57 |
robin_watts_mac | I'm bashing on the java stuff at the moment, so I'm happy to wait. Need to try to reset my body clock. | 15:41.52 |
mvrhel_laptop | ok sounds good | 15:42.13 |
chrisl | Hmm, Virgin Atlantic's e-tickets seem to have exploded in size - I don't remember it being 9 pages the last time! | 15:42.28 |
robin_watts_mac | oh, the marvels of JNI... | 15:42.28 |
| chrisl: At least you got one :( | 15:42.57 |
| All you need is the 6 character code. I just texted it to myself :) | 15:43.15 |
chrisl | robin_watts_mac: that was the e-ticket, tne boarding pass is still just a single page | 15:43.42 |
robin_watts_mac | Yeah, the e-ticket (which I didn't get updated versions of in the mail) is just 6 characters. | 15:44.21 |
chrisl | Right, yeh - I think I'll just pull the mail onto my phone the next time. | 15:45.14 |
| Anyway, all checked-in - should pack at some point, too! | 15:45.42 |
alexcher | Who knows if all cinfo->comp_bits can be equal 0 ? | 15:50.15 |
chrisl | So, are my code changes causing problems, or is the cluster having a tantrum, I wonder...... | 15:59.08 |
kens | Is there a problem ? I only just reconnected, having my machine on the network was cauisng a slowdown copying locally for some reason | 15:59.45 |
chrisl | I think it's on its second attempt, and five nodes have "went down, rerunning" and one says "Abort command received" | 16:00.50 |
kens | :-( | 16:00.59 |
henrys | all at miles' office pretty bad weather there I wonder if that's the problem. | 16:02.42 |
kens | Hm, Robin did say it was raining a lot | 16:03.04 |
chrisl | I was just going to say I couldn't remember where the nodes are.... I guess if this run completes, we'll know | 16:03.32 |
robin_watts_mac | kens, chrisl: Virgin PE seats have seatback power. Apple do a (oddly for them) very reasonably priced cable that ran my laptop the whole way (13 quid or something). | 16:07.52 |
chrisl | robin_watts_mac: what type of connector is it? | 16:08.27 |
kens | Thanks RObin but I don't think I could work for the whole flight.... | 16:08.28 |
| Where did you buy the cable from ? | 16:08.50 |
kens | doesn't have a Mac.... | 16:08.59 |
robin_watts_mac | chrisl: oh, it's a magsafe apple connector. | 16:09.00 |
chrisl | On the seat, I mean | 16:09.17 |
robin_watts_mac | but it's possible that other people may also sell them with different connectors. | 16:09.24 |
| oh, I see. | 16:09.27 |
| It's not the car lighter or mains ones. | 16:09.44 |
| EmPower I think. | 16:11.04 |
kens | need to go make a dead tree copy of my boardcing pass ane e-ticket, brb (btw my e-ticket is not 9 pages) | 16:11.05 |
robin_watts_mac | heads t obreakfast. | 16:26.19 |
mvrhel_laptop | robin_watts_mac: you there? | 17:21.05 |
robin_watts_mac | I am. | 17:21.11 |
mvrhel_laptop | Just got a call from Miles. He is going to be here in a bit | 17:21.16 |
robin_watts_mac | ah, ok. | 17:21.21 |
| So I'll get ready now. | 17:21.36 |
mvrhel_laptop | yes . me to | 17:23.17 |
paulgardiner | robin_watts_mac: I've added the external link handling (I think). It's pushed to my repo, should you get a chance to take a look. What I've done should be ok to push to master because the uri links aren't actually followed. | 17:25.12 |
robin_watts_mac | paulgardiner: Fab! | 17:25.44 |
kens | OK I'm heading off, I'll see you all tomorrow, one way or another | 17:37.11 |
robin_watts_mac | paulgardiner: That looks great to me. | 17:41.35 |
| You don't cope with GotoR links in there, right? | 17:41.46 |
| mvrhel_laptop: So you'll tell me when Miles gets here then? | 17:45.44 |
mvrhel_laptop | robin_watts_mac: yes. he is going to call me when he is about 10 minutes out. Apparently we need to help him unload the car. | 17:46.26 |
robin_watts_mac | Ah, ok. | 17:46.36 |
mvrhel_laptop | Scott was going to be here for the meeting as he had his flight changed but it was delayed so now he is going to miss the meeting | 17:46.49 |
| robin_watts_mac: ok miles is 5 minutes out. heading down now | 17:51.36 |
robin_watts_mac | ok. | 17:51.42 |
paulgardiner | robin_watts_mac: no. I should check his example file to see what sort of links he has. Still, now I have the architecture, I can add further cases fairly easily. Probably worth commiting what I have so far if you're happy with it. | 18:15.21 |
| robin_watts_mac: looks like they are all URI links | 18:21.14 |
henrys | chrisl:is there some killer feature in open wrt that you really like or just like or are you just playing around with it? | 18:52.43 |
chrisl | henrys: I wouldn't say it's a "killer" feature, but I wanted DNS on my LAN, without running a separate server - I couldn't find a router that advertised that | 19:10.03 |
henrys | yes I need that too. Most importantly for the network printer, I've been using tomato - geez I thought that would be standard in the regular firmware by now, maybe not | 19:12.04 |
| anyway I went with your recommendation. | 19:12.28 |
chrisl | Well, it may well be common, but I couldn't find any routers that said it in the spec - and none of the netgears I've had did DNS for the LAN. But I got sick of remember IP addresses! | 19:13.28 |
henrys | I don't mind IP addresses, what I like about tomato is you can associate a mac address with an IP, so they don't change all the time. | 19:14.42 |
| but I guess if you run full blown DNS you don't need that at all. | 19:16.08 |
chrisl | It's not really full blown DNS, it uses DNSMASQ for DHCP and DNS, for WAN lookups it just forwards to the upstream DNS, but has hostnames for all the dhcp leases (if the client is setup correctly) | 19:17.28 |
| henrys: I started here: http://wiki.openwrt.org/toh/netgear/wndr3800 | 19:17.52 |
henrys | thanks | 19:19.23 |
| I'll jump off that cliff when I return from california and the router has arrived. | 19:19.46 |
chrisl | Well, I'd assumed it would be a more involved setup than the stock "consumer" firmware, so I set aside a full day (with a half day spare) to get it all going - I was done in about 45 minutes. | 19:21.01 |
| henrys: Oh, two things: apparently, *some* of the netgear firware's have a firmware validation "feature", so will only reflash with a genuine netgear image (so not Open-WRT). The solution is to first "downgrade" to netgear's "1.0.0.16" firmware, then reflash with Open-WRT. Having said that, mine came with recent firmware, and flashed Open-WRT quite happily. | 19:24.50 |
| Second thing is that after reflashing it tells you to keep the router powered up for at least ten minutes - make sure you do that, it guarantees that the settings get updated in nvram before you reboot. | 19:25.52 |
henrys | thanks chrisl | 19:32.27 |
chrisl | np | 19:32.33 |
| Right, I'm heading off - see you all in SF! | 19:37.31 |
robin_watts_mac | back. | 21:18.25 |
Seq | Hello. Does anybody know a good way to get a page count from a PDF document using ghostscript? I'm currently looking at using the toolbin/pdf_info.ps grepping for "^Page" and piping to wc -l, but that seems hackish. | 21:28.10 |
| Previously used pdfinfo (from xpdf) but it often produces a different page count than what ghostscript provides -- most notably, it says any "portfolio" is only 1 page, whereas ghostscript flattens the portfolio and could output many pages. | 21:30.15 |
robin_watts_mac | There is a pdfinfo.ps tool with gs. | 21:30.52 |
henrys | robin_watts_mac: how'd it go? | 21:52.10 |
robin_watts_mac | well, I think. | 21:52.48 |
| They seemed genuinely interested, and impressed with the demo. | 21:53.19 |
mvrhel_laptop | yes. the demo went very well | 21:53.53 |
henrys | great news guys | 21:53.59 |
mvrhel_laptop | they want to have the apk to play around with | 21:54.14 |
Seq | robin_watts_mac, That's what I'm testing now, but I've got to grep the output of pdf_info.ps, for "^Page" and pipe through wc -l. I was just curious if there was a better way. | 21:59.21 |
henrys_ | sorry wireless at the coffee shop | 22:00.14 |
sebras | robin_watts_mac: oh no. | 22:01.03 |
| robin_watts_mac: I found another android bug. :( | 22:01.17 |
robin_watts_mac | sebras: oh. | 22:01.40 |
sebras | robin_watts_mac: I compiled HEAD just now. | 22:02.04 |
| basically zoom in full on a multipage text pdf. | 22:02.18 |
| then click search | 22:02.22 |
| enter a common phrase. | 22:02.29 |
| click next, next, next, rapidly. | 22:02.36 |
| eventually you'll end up on a page full zoomed in but the page is never redrawn at the correct DPI. :/ | 22:03.07 |
robin_watts_mac | Seq: Well, if you're up for postscript programming you can take pdfinfo.ps to bits. | 22:03.12 |
sebras | though if you scroll a teensy bit, everything will be redrawn of course. | 22:03.31 |
Seq | robin_watts_mac, I may end up heading toward that route, although I've never looked at postscript before. | 22:04.19 |
robin_watts_mac | sebras: hmm. Well, as long as it doesn't crash. | 22:05.01 |
sebras | robin_watts_mac: haven't managed to kill it so far. should I try or do you want to retain some peace of mind while in .us? ;) | 22:05.42 |
robin_watts_mac | sebras: Feel free. I can send the letterbomb when I return :) | 22:06.18 |
sebras | robin_watts_mac: :) | 22:06.54 |
mvrhel_laptop | robin_watts_mac: after digging a bit more, reading and thinking about it, the knock_out non isolated case is a much bigger pain than the other modes. This is due to the fact that I can have two objects that overlap one another. Each to be drawn with different blend modes. And if in a knockout non-isolated group, the overlapping one must be drawn against the backdrop with its blend mode... | 22:17.48 |
| ...as if the other one had not occured | 22:17.50 |
| the mixing of blendmodes makes it not possible to do it in whole group compositing action after doing a bunch of knockout drawings | 22:18.45 |
robin_watts_mac | so... it'll take tomorrow as well as today then? | 22:19.02 |
mvrhel_laptop | ha | 22:19.08 |
| the cleanest approach is to have access to the back drop always to ignore anything that was previously drawn | 22:20.23 |
| in essence, use the parent buffer | 22:20.41 |
| as I drawer in the current buffer | 22:21.02 |
| draw | 22:21.06 |
robin_watts_mac | is it a single buffer, or a whole mess of buffers? | 22:23.16 |
mvrhel_laptop | So there is the parent buffer from which we initialize our backdrop | 22:24.02 |
| the backdrop being the new buffer | 22:24.20 |
| which we currently draw in | 22:24.25 |
| however, we have to keep a clean backdrop or be able do undo things that were previously drawn | 22:24.52 |
| until we close the group | 22:25.03 |
| so, what I would propose is that I access the parent buffer for the background value always as we do the fill of our current buffer | 22:25.54 |
robin_watts_mac | so you read from one place and write to another ? | 22:26.19 |
mvrhel_laptop | yes | 22:26.24 |
robin_watts_mac | how huge an architectural change is that? | 22:26.37 |
mvrhel_laptop | the only issue I see is that there is currently not a pointer back to the parent buffer | 22:27.19 |
| alternatively, I would need to add another buffer | 22:27.31 |
robin_watts_mac | All the blending functions assume that they are currently doing: | 22:27.43 |
mvrhel_laptop | which is my clean backdrop | 22:27.48 |
robin_watts_mac | *s = COMBINE(s, t) | 22:28.07 |
| *s = COMBINE(*s, *t) even | 22:28.18 |
| So you'd need to change that to be: *s = COMBINE(*t, *u) | 22:29.00 |
mvrhel_laptop | right | 22:29.11 |
robin_watts_mac | where most of the time s==t ? | 22:29.12 |
mvrhel_laptop | if I understand your notation | 22:29.30 |
robin_watts_mac | ok, so it means the compiler is juggling another value. | 22:29.34 |
| several more values by the time you're allowing for increments/spans etc. | 22:29.58 |
| but at least (most of the time) we won't be dragging another buffer into/out of cache. | 22:30.14 |
mvrhel_laptop | So, you agree that being able to access the parent buffer to get the current backdrop sounds reasonable | 22:31.48 |
| not sure if I am being super clear. Maybe we can chat in person | 22:33.11 |
| I guess we are meeting in a few to put up the booth | 22:33.43 |
robin_watts_mac | mvrhel_laptop: Changing the blend functions to be a function of dst = src1 + src2 rather than dst = dst + src seems reasonable, yes. | 22:34.50 |
mvrhel_laptop | oh. I see what you are saying. ok. luckily the low level blend functions are already a function of src1 and src2. They just always get the current buffer value and the new value to be drawn in. I just need to get it the right values, which means accessing the parent buffer for one of the values in the knockout, nonisolated case | 22:38.01 |
robin_watts_mac | ah, right. | 22:38.19 |
| So the per pixel functions don't need to change. | 22:38.30 |
mvrhel_laptop | right | 22:38.33 |
robin_watts_mac | Just the setup to call them. | 22:38.35 |
mvrhel_laptop | exactly | 22:38.41 |
robin_watts_mac | Right, that's a much less invasive change then, great. | 22:38.49 |
mvrhel_laptop | yes | 22:38.53 |
| at least I hope..... :) | 22:39.03 |
robin_watts_mac | What you suggest sounds entirely plausible to the limited extent of my understanding. | 22:39.23 |
mvrhel_laptop | ok. thanks for listening and the feedback. | 22:39.43 |
robin_watts_mac | np. | 22:39.59 |
| Forward 1 day (to 2012/12/06)>>> | |