| <<<Back 1 day (to 2014/07/28) | 2014/07/29 |
jogux | does anyone know if artifex have ever licsensed a wingdings or wingdings 2 font from anywhere? | 11:29.51 |
kens | Isn't the URW font a wingdings replacement ? | 11:30.22 |
kens | goes off to look | 11:30.27 |
tor8 | jogux: AFAIK we only have Dingbats | 11:30.28 |
kens | Ah Dingbats, right | 11:30.36 |
jogux | ah right, thanks. | 11:30.57 |
henrys | almost meeting time | 14:28.47 |
mattchz | http://developers.slashdot.org/story/14/07/29/1234229/put-your-code-in-the-swamp-dhs-sponsors-online-open-source-code-testing | 14:31.04 |
| Wonder if thatâs relevant/useful for mupdf/gs | 14:31.12 |
henrys | about the free mupdf on android we can put a free apk on the mupdf site and but sell on the app store next release. Thatâs what Miles will agree to. | 14:31.26 |
mvrhel_laptop | morning | 14:32.02 |
tor8 | henrys: I'm still not convinced it's a good idea to start selling the app, but I'm happy to give the finger to walled gardens of any sort | 14:32.39 |
henrys | we can also have a free ios version on the mupdf site, of course thatâs not likely to be used | 14:32.42 |
tor8 | henrys: that's not possible, though. the only way to install stuff on ios is via the app store. | 14:33.00 |
henrys | tor8: no you can jailbreak the phone | 14:33.17 |
| ipad | 14:33.19 |
| obviously we expect very few takers of that. | 14:33.50 |
tor8 | I thought that was still illegal in certain jurisdictions | 14:33.53 |
kens | mattchz is that more than just static analysis ? We've used those before.... | 14:34.17 |
henrys | tor8: so? | 14:34.18 |
mattchz | kens: Iâve not looked in detail, but I think itâs static analysis tools. | 14:34.32 |
henrys | tor8: I donât think there is an abetting issue, is that what you mean? | 14:34.45 |
tor8 | henrys: we may not want to advice people to jailbreak their devices | 14:34.46 |
kens | It seemed ot imply that. We've used thse in the past, GS tends to confuse them quite a bit. | 14:35.06 |
tor8 | but I'm more interested in seeing the download and purchase figures from mupdf on the ios store since we started charging for it | 14:35.18 |
rayjj | we could probably say that if your phone is unlocked, then you can use this free version... | 14:35.35 |
henrys | I donât feel strongly about it tor8 we can just put up the android apk | 14:35.44 |
| for the record I am completely against all of this. Miles wants to try it for 6 months | 14:36.26 |
chrisl | henrys: any indication why Miles suddenly thinks we'll money out of it? | 14:36.55 |
rayjj | mvrhel_laptop: I think I found the bug that was causing the problem with the focus jumping in gsview. | 14:37.05 |
marcosw | rayjj: in cell phone language unlocked means not tied to a carrier, jailbroken is what allows you to install apps not through the app store. | 14:37.22 |
tor8 | yeah. I fear that it will harm us and our reputation more than the income would be worth | 14:37.22 |
mvrhel_laptop | rayjj: oh great. | 14:37.24 |
henrys | chrisl: miles is a very optimistic guy - I assume weâll make bupkiss in 6 months and we can go back to business as usual. | 14:37.54 |
rayjj | mvrhel_laptop: that or it went away. It was hard to reproduce | 14:38.17 |
henrys | I did present everyoneâs argument feel free to write to him if you feel strongly about it. | 14:38.50 |
chrisl | henrys: I'm just bemused by the change - I'm guessing that with SOT on the app store, Miles reckons app stores are a money spinner now..... | 14:38.59 |
rayjj | henrys: that sounds like a plan. Then we can announce how wonderful we are because we decided to give away this valuable ($0.99) app for free :-) | 14:39.33 |
paulgardiner | rayjj: did you get a chance to look at the tmp-file fix yet? | 14:40.45 |
henrys | then it will have all the negative reviews and be free, and people will think wow this must really suck if folks are complaining about a free app ;-) | 14:41.09 |
| anyway onward marcosw did you see that potential customer from scott canât download (chrisl) | 14:42.28 |
| this seems to be a recurring theme, now | 14:42.40 |
paulgardiner | The thing that worries me is that we don't really know how good the free app is at pulling in licencees, so however much the app store makes, we wont know if it's an overall win or not | 14:42.57 |
chrisl | henrys: I called godday yesterday, was on hold for 20 minutes, and gave up..... will try again at some point | 14:43.08 |
marcosw | the guy from australia? I offered to mail him a dvd-rom. | 14:43.09 |
chrisl | Or even GoDaddy | 14:43.24 |
henrys | chrisl: okay I wonder if we should start looking for an alternative | 14:44.08 |
marcosw | henrys: every time you release a new version on the ios store the reviews get reset to zero, so as long as we do that when we change the price it won't be an issue. | 14:44.19 |
rayjj | paulgardiner: no, I've mostly been wrapped around the axle with cust 532's problem(s). When I get fed up, or right before stopping for the day I play with gsview for a few minutes | 14:44.40 |
henrys | marcosw: oh well thatâs good | 14:44.44 |
chrisl | henrys: I've no idea where to look - reasonably priced, unlimited downloads are not common these days :-( | 14:44.47 |
henrys | chrisl: weâll have to wait for marcosw to get fiber | 14:45.07 |
| ;-) | 14:45.18 |
paulgardiner | rayjj: ah. My sympathies. | 14:45.27 |
marcosw | I'll just move to one of the google fiber cities :-) | 14:45.40 |
chrisl | henrys: we could look into if amazon had a viable cloud "plan" | 14:45.56 |
henrys | marcosw: it is rather surprising you donât have something there, if anyplace would have the demand I would think it would be there | 14:46.17 |
marcosw | Austin, TX seems better than Provo, UT or Kansas City. | 14:46.22 |
| San Jose is on the list, not sure if that includes Fremont (sometimes when people say San Jose they mean Silicon Valley). | 14:46.55 |
| apparently not. as far as google is concerned San Jose means San Jose, Santa Clara, Mountain View, Sunnyvale, and Palo Alto. | 14:47.53 |
mvrhel_laptop | where is the list.. | 14:47.59 |
marcosw | https://fiber.google.com/newcities/ | 14:48.05 |
henrys | tor8: jogux just reported 10.00 a d ay | 14:48.09 |
| on skype | 14:48.26 |
marcosw | mvrhel_laptop: Portland is the closest city to Seattle. | 14:48.28 |
kens | henrys 10.00 what ? | 14:48.47 |
henrys | tor8: well 10 copies a day | 14:48.55 |
kens | ah | 14:48.59 |
henrys | kens: ios sales | 14:49.02 |
jogux | marcosw: the reviews don't get entirely reset for the new version btw, you can always click through to see the 'all versions' reviews. | 14:49.08 |
marcosw | Nashville is on the list, my brother in law lives there. | 14:49.16 |
| jogux: didn't realize that. I always see apps requesting to be re-reviewed when they release a new version. I wonder how often people look at the 'all verison' reviews. Presumably they don't get shown in the search results on the app store. | 14:50.25 |
jogux | it might be worth just having a second copy of the website somewhere we can direct people to? albeit I have no idea how godaddy could mess things up enough that people couldn't download | 14:50.50 |
tor8 | henrys: jogux: what was the average download rate per day while we were free? | 14:51.15 |
jogux | marcosw: yeah, the dynamics aren't entirely clear to me. probably people do when a new version is just released and there's not many reviews. | 14:51.19 |
| tor8: just a sec | 14:51.22 |
rayjj | mvrhel_laptop: oops. Nope. The bad news is that I can still get gsview to get messed up. The good news is that I can now easily reproduce the problem. I'll put it in the bug 695389 | 14:51.42 |
marcosw | I did try downloading all of the AGPL versions and had no problems. Do we know if the problem is geographic, transient, or both? | 14:51.46 |
mvrhel_laptop | rayjj: ok. I will have a look at these this week. | 14:52.02 |
jogux | tor8: not sure I have the full data, but 110 per day is probably a reasonable average for the last month. | 14:52.09 |
henrys | marcosw: I thought last time they were throttling us. | 14:52.20 |
jogux | tor8: 2800 units for July so far. | 14:52.44 |
| with 30 of those being paid. | 14:52.51 |
tor8 | jogux: that's more than I expected :) I recall it hovering around 30 downloads/day for months on end in the beginning | 14:52.58 |
chrisl | henrys, marcosw: I got the dowonloads to fail by repeatedly downloading - about 1 in 5 was incomplete. But not every one over 5..... | 14:53.28 |
henrys | tor8: so miles is eaget to get an android paid app out of course. | 14:53.35 |
jogux | does anyone know how much bandwidth the downloads site uses? | 14:53.45 |
marcosw | henrys: throttling suggests a reduction in speed, not failure. | 14:53.50 |
tor8 | so that's 90% of potential users we could impress gone... | 14:53.57 |
marcosw | jogux: I can tell you how much we used to use when we hosted downloads on Amazon ec2. But that was a while ago... | 14:54.21 |
tor8 | 75% of who would try the app once then uninstall it, of course :) | 14:54.24 |
chrisl | henrys: the last time had these problems, it was with the previous hosting co we used | 14:54.26 |
jogux | tor8: :-) | 14:54.34 |
henrys | jogux: for ghostscript, thatâs fairly heavy traffic but I havenât checked it recently. | 14:54.39 |
jogux | tor8: we have more figures now actually, use a sec. | 14:54.42 |
chrisl | marcosw: I thought we had stats from the GoDaddy site | 14:55.10 |
jogux | tor8: crashlytics reports that, for the latest ios build, there's around 400 daily active users. | 14:55.36 |
henrys | chrisl: oh I thought we had problems with these guys. Youâre sure? | 14:55.42 |
marcosw | chrisl: possibly, but I can't access those. | 14:55.56 |
chrisl | henrys: I don't remember having problems with GoDaddy before - I could be wrong | 14:56.12 |
henrys | mvrhel_laptop: howâs gsview? | 14:56.25 |
tor8 | jogux: how big was the window between crashlytics being added and becoming paid? | 14:56.37 |
marcosw | okay...firefox crashed when I tried open an old amazon aws invoice. | 14:56.45 |
tor8 | though I guess everybody with an old copy would get automatically upgraded so probably not relevant | 14:57.02 |
mvrhel_laptop | henrys: I have rayjj up and running/building with gsview. he has found some issues that we are going to work on on the side | 14:57.03 |
jogux | tor8: the crashlytics figures include all the existing users that upgraded for free, so I don't think you can deducate anything about paid users from these figures. | 14:57.05 |
henrys | mvrhel_laptop: for printing you should add -dNOCACHE to the process arguments. | 14:57.21 |
jogux | s/deducate/deduce/ | 14:57.21 |
tor8 | jogux: yeah, just realized that. my brain is a bit sluggish in this 30 degree weather. | 14:57.32 |
mvrhel_laptop | henrys: oh ok. I will do that | 14:57.37 |
henrys | anything else for the meeting this week? | 14:57.43 |
chrisl | henrys: the sales (at) artifex.com mails going missing - seems to be quite a bit of that recently | 14:58.22 |
| I wondered if we should have it go to a mailing list so more people have visibility of it | 14:58.53 |
henrys | chrisl: yes I spoke to scott about that | 14:59.15 |
jogux | chrisl: I kinda of hate to say it, but feeding it into a CRM/ticket system might help? | 14:59.19 |
tor8 | chrisl: given how badly written a lot of the questions we get are, I wouldn't be surprised if a lot of them are hiding in Scott's spam box | 14:59.20 |
rayjj | that or Scott just hits "delete" | 14:59.52 |
chrisl | Well, it's worrying because I don't imagine the ones that come on here are the only ones not getting a reply | 14:59.54 |
rayjj | at least one that I talked to him about he "found" when I asked him about it | 15:00.35 |
henrys | scott seems to think it is not an issue. Iâll add it to the agenda. | 15:00.54 |
marcosw | the last month I have data for (in 2011) had us transfering 1.6 TB. That's total from amazon to the internets, it went down to 300 GB when we stopped hosting the downloads, so (marcosw get's out a calculator), it appears our downloads of Ghostscript were 1.3 TB/month. | 15:01.15 |
rayjj | I don't know how much spam comes to that email. That may be the problem | 15:01.20 |
henrys | Iâll put my name on the sales group if you want. I donât miss many. | 15:01.56 |
rayjj | henrys: two sets of eyes are better than one, but how will you know if Scott responds unless he cc's you ? | 15:02.34 |
jogux | marcosw: hm. well, my 10quid/month VPS comes with 400G per month included, and it's one of the less generous ones. | 15:02.36 |
| daft question, but couldn't downloads of ghostscript etc be hosted on sourceforge or similar? | 15:02.53 |
| at least as a backup? | 15:02.59 |
henrys | rayjj: Iâll have him cc me. | 15:03.07 |
rayjj | jogux: we tried that. sf got to be a pain | 15:03.20 |
jogux | rayjj: ah :-( | 15:03.31 |
marcosw | amazon charges $0.12/Gig so we'd be paying ~$150 with them. | 15:03.33 |
henrys | or would you like to be on it? volunteers for this fun duty? | 15:03.38 |
rayjj | jogux: but maybe some other outfit | 15:03.45 |
chrisl | jogux: sourceforge were pushing the ads, and some other crap, and google code stopped allowing "simple" file downloads | 15:04.12 |
rayjj | henrys: I already get enough spam | 15:04.14 |
mvrhel_laptop | henrys: I have to leave in 10 minutes today. I will be back in about an hour though | 15:04.34 |
jogux | henrys: I'd really suggest some kind of CRM for both support & sales tbh. | 15:04.36 |
marcosw | google's cloud charges within 10% of amazon prices (or did when they were first announced, haven't looked lately). Same with HP. | 15:04.41 |
jogux | chrisl: ah :-( | 15:04.43 |
| azure was cheap for a bit. but amazon's prices have come down a lot in the last year. | 15:04.58 |
chrisl | Oh, and sourceforge also went through a phase of incomplete downloads (with no error given)....... | 15:06.17 |
marcosw | hp and google both charge $0.12/GB as well, reminds me of airline pricing. | 15:06.38 |
henrys | jogux: I doubt I could foist that on Scott | 15:07.36 |
chrisl | I'll try calling GoDaddy again, but I doubt I'll get anywhere, as Robin setup and paid for the account | 15:07.43 |
henrys | chrisl: I thought he gave all the access info to ron. No? | 15:08.10 |
jogux | henrys: true :-S unless you found one that actually made his life easier somehow. | 15:08.21 |
chrisl | henrys: username and password, sure, but there's stuff about pin numbers which I've no idea about | 15:08.40 |
jogux | I could mirror stuff onto our london VPS as somewhere you could point people at if they say the ghostscript.com downloads fail, we've a few 100G per month going unused. | 15:09.26 |
| it's 0.10 ukp per gigabyte for overage though, so quickly gets expensive over the included 400G. | 15:10.16 |
henrys | chrisl: I thought he changed everything over. | 15:10.39 |
| billing all that. | 15:10.47 |
mvrhel_laptop | bbiab | 15:10.52 |
chrisl | henrys: payment information is still Robin's name, address etc | 15:11.20 |
| Well, as of the last payment | 15:11.39 |
henrys | if not he really should. heâll get impale by a mountain goat in the 16000 feet in bolivia and weâll be screwed. | 15:11.48 |
chrisl | henrys: even if he has changed the details, it will probably be the last payment details that count. If I actually get through to them, I'll find out | 15:12.36 |
marcosw | I'd be more worrid that an easter island head falls on him. impaled by a goat isn't likely to be fatal. | 15:13.09 |
chrisl | rayjj: could you review a commit for me? http://git.ghostscript.com/?p=user/chrisl/ghostpdl.git;a=commitdiff;h=62d35980 | 15:14.46 |
henrys | rayjj: you set up scottâs mail, how does he see spam google collects? Is the spam folder downloaded? | 15:17.22 |
| tor8: 30 degrees Iâm ecstatic we are getting a break in the heat and my race next sunday will be *only* 30 degrees, itâs a miracle | 15:20.02 |
marcosw | I'm off to uni, be back online in a while. | 15:20.29 |
tor8 | henrys: it's all relative to what you're used to :) | 15:20.38 |
henrys | tor8: yup, I guess that is warm for sweeden | 15:21.02 |
tor8 | the construction workers outside have been making so much noise I've had to keep the windows closed :( | 15:21.16 |
kens | is pleased the temperature here has dropped below 25 degrees again | 15:22.08 |
rayjj | chrisl: looking... | 15:22.37 |
| henrys: I don't think google's spam is downloaded at all. It just downloads the inbox. Anything filtered on gmail out of the inbox isn't considered for POP download | 15:23.38 |
chrisl | You can see google's spam folder (junk mail?) if you use imap, and have your client subscribe to that folder | 15:25.03 |
henrys | rayjj: thatâs what I expected, I asked him to use gmail directly on the web I think it would be best for him, but he didnât want to do that. | 15:25.47 |
rayjj | henrys: gmail on the web is SLOW for sorting, IMHO | 15:28.33 |
henrys | rayjj:I use it and keep a syncâd copy for fast searches and such | 15:29.32 |
rayjj | chrisl: you don't need c_alone, IMHO. Where you have !imem->cc.c_alone && just use !((obj_header_t *)(imem->cc.cbase).o_alone) If the chunk has an single object, it will always be first (at cbase, right ? | 15:39.26 |
chrisl | rayjj: I thought strings would be different - that's why I opted for the new flag | 15:40.41 |
rayjj | chrisl: and there is a missing word in the log message: "The problem that occurs is when we free an o_alone object, we immediately the" s/b "The problem that occurs is when we free an o_alone object, we immediately free the" | 15:40.58 |
chrisl | rayjj: okay, I'll double check the strings case when I get off the phone with GoDaddy. | 15:42.20 |
rayjj | chrisl: TBH, I've never dug into the mess of string allocations | 15:42.24 |
chrisl | rayjj: IIRC, strings are allocated top down - so in this case, it will depend on where the padding is | 15:42.58 |
rayjj | chrisl: but are strings ever put into a chunk as o_alone ? Or I didn't understand you | 15:44.07 |
chrisl | Yes, immovable strings are o_alone | 15:44.27 |
rayjj | chrisl: Oh, ok. So the "trick" to use cbase as an obj_header_t wouldn't be valid. | 15:45.12 |
chrisl | No, if you look in i_alloc_string_immovable(), the string is at the top of the block, so any padding/rounding will get in the way | 15:45.58 |
rayjj | chrisl: and cbase doesn't ever point to an obj_header. Your patch is fine (with the log message change). Thanks for the discussion. | 15:46.59 |
chrisl | rayjj: thanks - have to say, that was a pretty nasty one to track down! | 15:47.35 |
rayjj | chrisl: I bet! | 15:47.56 |
| I have to go now (take car in) | 15:52.23 |
chrisl | henrys: okay, so GoDaddy have reproduced the problem (or a problem) with our downloads, and they are going to investigate | 15:54.28 |
rayjj | chrisl: thanks. I'm amazed that they actually were able to find a problem | 15:54.58 |
henrys | chrisl: good news | 15:55.22 |
chrisl | rayjj: I was a bit surprised, too. We'll see if it bears any fruit, though..... I'll check the downloads tomorrow, and call again if there's still a problem | 15:56.00 |
rayjj | mvrhel: (for the logs). Should I open a bug for this "The icon in the task tray is the 'generic' icon, not the ghostie icon as with the gsview image window" | 15:56.13 |
| mvrhel: oops. That's only when running under the debugger. When invoked dierctly, it's OK. NM | 15:58.08 |
rayjj | is glad I didn't open a bug :-) | 15:58.19 |
chrisl | I wish people wouldn't repeatedly use the same name for their test jobs for different bugs :-( | 15:58.32 |
rayjj | chrisl: I always save as Bug______.pdf so that the names may not match what they have in the bug, but at least I can associate them | 15:59.28 |
henrys | chrisl: how does that memory problem just get noticed now? | 15:59.37 |
rayjj | henrys: I'm just guessing, but resizing objects is pretty rare | 16:00.05 |
chrisl | Not in this case, the object in question was the raster buffer for buffered x11 output...... | 16:00.38 |
| So it gets resized quite a lot | 16:00.49 |
rayjj | chrisl: right -- every time the page size changes | 16:01.11 |
| and it's big enough to be o_alone | 16:01.30 |
| ok, really going now... | 16:02.04 |
henrys | chrisl: I wonder if we lack tests for that ⦠| 16:02.32 |
| other problems may lurk | 16:02.44 |
chrisl | henrys: historically, I don't think we have collected many chronically damaged PDF files - and also then, having identified a broken PDF, carried on regardless in the hope nothing horrid will go wrong later on | 16:03.42 |
| henrys: the changes to have the PDF interpreter try to carry on in the face of broken input rather breaks the rules of the Postscript world. | 16:04.51 |
henrys | chrisl: I was thinking for real world use cases where the framebuffer is resized during the course of a job - how well tested that is. | 16:05.34 |
chrisl | henrys: probably rather well - an awful lot of our PDF tests use page sizes that are not our default | 16:06.16 |
| BTW, with most of these fuzzing PDFs, if I run with PDFSTOPONERROR, no segfault (or whatever) occurs | 16:07.37 |
kens | is unsurprised. I feel that should be the default | 16:13.43 |
| Night all | 16:24.05 |
henrys | marcosw: ugh I donât think scott should have referred that to support, didnât we agree on bugzilla? and PCL is not GPL, it is limited by the AFPL of itâs fonts. | 16:53.58 |
chrisl | henrys: the PCL code is AGPL | 16:58.49 |
henrys | thatâs moot the point is the fonts are AFPL - not for commercial use | 16:59.34 |
| alladin license | 16:59.50 |
chrisl | True, yes. And either way, support is the wrong place | 17:00.38 |
henrys | though I should get in the habit of saying AGPL | 17:00.42 |
| yup âbroken recordâ mode doesnât see to work Iâll have to try something else | 17:02.07 |
chrisl | I guess the only thing might be if Scott considers them a prospect for a support contract..... | 17:03.18 |
marcosw_ | anyone have any idea why this command line: "mudraw -o test%d.pdf input.pdf" generates a output file with the name test%d.pdf instead of replacing %d with the page number? | 17:16.28 |
henrys | marcosw_: I updated http://bugs.ghostscript.com/show_bug.cgi?id=695374 with a question, just wanted to make sure you saw it. | 17:19.36 |
marcosw__ | henrys: I've updated http://bugs.ghostscript.com/show_bug.cgi?id=695374 | 17:33.57 |
henrys | marcosw__: thanks I did the timings also but best to have them relative to your other numbers | 17:35.13 |
marcosw__ | there is a lot of noise in the data, I wouldn't trust any differences <10%. | 17:35.58 |
d1gital | is there any character aside from unmatched parens that should not be allowed in a ()-enclosed string? | 20:39.34 |
rayjj | d1gital: If you are talking PostScript, then, yes, there are other characters that need to be escaped | 22:05.20 |
| d1gital: and if you are sending the PS to a printer, there may be other characters that are problematic if the channel is not binary capable | 22:07.50 |
| In PDF's, within a literal string, the backslash (\) is used as an escape character for various purposes, such as to include newline characters, non-printing ASCII characters, unbalanced parentheses, or the backslash character itself in the string. The character immediately following the backslash determines its precise interpretation. If not one of n, r, t, b, f, (, ), \ an EOL or 3 octal... | 22:13.31 |
| ...digits, then the \ is ignored. | 22:13.33 |
| Forward 1 day (to 2014/07/30)>>> | |