| <<<Back 1 day (to 2015/05/18) | 20150519 |
kens | I can't find the purported customer from the latest email to support (pdfwrite crashing email), but I have a feeling they are (or were) a customer. Anyone have any idea what they might be known as ? | 09:16.24 |
chrisl | They are a customer..... | 09:19.28 |
kens | I thought they were,but I can't find them in the list.... | 09:19.44 |
chrisl | Miles mentioned them in his "new customer" review at the London staff meeting | 09:21.40 |
| His agenda mail sent on 05/12/2014 | 09:21.59 |
kens | Ah! well that explains why I remember them, but not why they aren't on the customer list.... | 09:22.13 |
| Or perhaps I just can't find them, I sem to havre tha tproblem a lot | 09:22.39 |
chrisl | I don't see them under that name, anyway... | 09:22.56 |
kens | ThenI will send my email to support asking Marcos :-) | 09:23.14 |
| Hmm, the original request for a quote from tehm says they want to run on HP-UX, but that looks more like Windows to me.... | 09:37.48 |
| Ah, they are the drilling rigs people. I guess they are using GS to make PDF files of their documentation. | 09:39.51 |
kens | thnnks Hin-Tak rather missed the point of the bug being assigned to him :-) | 12:30.29 |
Robin_Watts | kens: Yes, I get the problem with my current code. | 12:32.10 |
| but then I may be a couple of weeks out of date. | 12:32.20 |
kens | Oh, because I don't. | 12:32.21 |
| I tested with the current HEAD of master and I cannot reproduce the problem there | 12:32.45 |
| I could with both 9.15 and 9.16 as I said, whch is why I presumed that one or more of Ray's changes had fixed it. | 12:33.13 |
Robin_Watts | ok, it's probably been fixed in the last few weeks then. | 12:33.50 |
kens | I should say Ray or Michael's changes.... | 12:33.53 |
tor8 | Robin_Watts: I'm adding user stylesheets to the EPUB reader, and I can't decide how/where to feed the stylesheet in | 13:12.10 |
Robin_Watts | tor8: ah. | 13:12.33 |
tor8 | one option is to set it on a document after it's been opened (or add another argument to the constructor) | 13:12.35 |
Robin_Watts | tor8: Something in the context? | 13:12.43 |
tor8 | another one is to set it in a global context | 13:12.44 |
Robin_Watts | The idea is to set user styles that apply to all documents loaded, right? | 13:13.20 |
| So the context seems the right place. | 13:13.27 |
tor8 | on the other hand, changing the stylesheet in the global context will make it harder to trigger a re-layout | 13:13.32 |
| yeah, at the moment I'm going to add a -U option to mudraw and mupdf | 13:13.45 |
| or it could pick it up from a config file like ~/.mupdf.css | 13:13.58 |
Robin_Watts | config file should merely inform the context. | 13:14.18 |
tor8 | but a viewer should be able to use the user stylesheet to change defaults like margins and font selections, was my thinking | 13:14.30 |
Robin_Watts | At the moment we have no 'list of all documents' in the context, do we? | 13:14.37 |
tor8 | we do not | 13:14.46 |
| fz_set_user_stylesheet(ctx, doc, css_text) is one option, and then I'd leave the management of all css to the client | 13:15.14 |
Robin_Watts | So we'd need to add that, and make a meta call (or a new call) when the context changes. | 13:15.20 |
tor8 | but it feels hackish, but similar to the way we now call fz_layout_document to trigger a reflowing | 13:15.40 |
Robin_Watts | tor8: AIUI, that would mean that you load the document, lay it out, then load the css and have to lay it out again. | 13:15.43 |
| i.e. there would always be 2 layouts. | 13:15.55 |
tor8 | in practice you'd have to re-parse the document when the user stylesheet changes | 13:16.04 |
Robin_Watts | tor8: Yes, but ideally you'd set the user stylesheets before initially loading the doc. | 13:16.25 |
| thus only one layout. | 13:16.33 |
| hence the style sheets being in the context feels right to me. | 13:16.46 |
tor8 | so setting the stylesheet in a separate function call irks me, and adding it to the constructor is awkward for all the non-epub formats | 13:16.58 |
Robin_Watts | How about web browsers? User style sheets are per browser, not per pane, right? | 13:17.04 |
tor8 | setting it in the context, and have the document pull it out of there sounds like a reasonable choice | 13:17.29 |
Robin_Watts | tor8: It's equivalent to setting the AA level, I feel. | 13:17.39 |
tor8 | and then leave it up to the client to trigger a relayout or reparse if it changes the user css | 13:17.51 |
Robin_Watts | We set that at the context level, and docs refer to it. | 13:17.55 |
| tor8: yeah. | 13:18.07 |
tor8 | Robin_Watts: okay, sounds reasonable. | 13:18.16 |
Robin_Watts | Based on my comprehensive 15 seconds thought, that sounds like the right course of action :) | 13:18.27 |
tor8 | Robin_Watts: I'm also thinking of adding font selection similarly | 13:18.41 |
| so having a stylesheet context sounds like a plan | 13:18.49 |
Robin_Watts | For additional fonts? yeah. | 13:18.54 |
tor8 | user_stylesheet | 13:18.56 |
| yeah. times isn't the prettiest font to use :( | 13:19.06 |
Robin_Watts | gotta grab lunch. bbs. | 13:19.56 |
tor8 | Robin_Watts: several commits on tor/master for consideration | 14:27.38 |
henrys | meeting time, happy Tuesday! | 14:30.10 |
mvrhel | morning | 14:30.14 |
fredross-perry | howdy | 14:31.00 |
ghostbot | Welcome to #ghostscript, the channel for Ghostscript and MuPDF. If you have a question, please ask it, don't ask to ask it. Do be prepared to wait for a reply as devs will check the logs and reply when they come on line. | 14:31.00 |
henrys | mvrhel: you first since you have to leave early. Anything going on with gsview? | 14:31.01 |
Robin_Watts | mvrhel has just tracked down a wierd memory leak that only manifested on windows 7 and earlier reported by a customer who was using a fragment of the gsview code. | 14:32.03 |
mvrhel | henrys: I am doing some fixes to avoid using the call to pdf_clean main and writing a script to make it easier for everyone to build the winrt projects | 14:32.09 |
henrys | mvrhel: It looks like the customer is taken care of, it was starting to look like we'd be getting near charging extra. | 14:32.11 |
mvrhel | henrys: Yes. I think he is all set for now | 14:32.30 |
| I hope | 14:32.32 |
henrys | mvrhel: I've been keeping an eye on the conversation for an NRE. We'll see if there is more. | 14:33.08 |
Robin_Watts | henrys: For a while I was convinced it was PEBCAK, but it turned out to be genuine. | 14:34.02 |
rayjj | mvrhel: are those fixes are relevant to gsview? (to stop memory leaks) | 14:34.05 |
mvrhel | rayjj: they are | 14:34.20 |
henrys | mvrhel: okay also fredross-perry is still available for part time mupdf - delegate! | 14:34.33 |
rayjj | so, it's not wasted on the single customer | 14:34.36 |
Robin_Watts | Memento has been tweaked so that it can check C++ allocations too. | 14:34.47 |
henrys | hard to believe you don't have something like valgrind on windows, I know there are tools you can pay for... | 14:35.40 |
mvrhel | yes. I am going to do a bit more testing with mememto and gsview to see if there are any other issues when doing other features | 14:35.53 |
henrys | fredross-perry: you've been mostly sot this week yes? | 14:36.11 |
mvrhel | I did run the code analysis in visual studio. that found a couple issues I was surprised to see | 14:36.15 |
Robin_Watts | MSVC has tools in it, but the final problem was specific to C#, and those tools don't work on < windows 8. | 14:36.25 |
rayjj | henrys: there are leak checking tools in VS 2013/15 | 14:36.31 |
fredross-perry | yes, sot. continuing with that unles thereâs a different priority. | 14:36.38 |
Robin_Watts | and this problem didn't exhibit on < windows 8... | 14:36.40 |
mvrhel | yes you need 8+ to use the memory tools and a few others | 14:36.56 |
rayjj | Robin_Watts: only on < win 8 | 14:36.59 |
Robin_Watts | rayjj: oops, yes. | 14:37.20 |
henrys | kens2: anything to report about the FirstPage/LastPage branch? | 14:38.08 |
rayjj | that's why mvrhel had so much trouble on win 7 -- it only leaked there and no tools | 14:38.33 |
mvrhel | yes. but now I have a full setup in win 7 for continued testing | 14:38.33 |
kens2 | Its basically working. I have now two devices (an object filtering one as well) to prove that they can co-exist. I'm working on Zoltan Biro's problem. Then I need to write up what I've done and make some docuementation. I'm planning to have it all done by the staff meeting | 14:39.01 |
mvrhel | nice | 14:39.16 |
henrys | I have been working on the combining fonts for the language switch build and indeed some of the metrics are different still researching. i want to do some tests on my HP printer comparing the metrics. | 14:39.26 |
| kens2: great | 14:39.57 |
kens2 | henrys save commendation till you see it :-) | 14:40.11 |
marcosw | henrys: so I can look forward to yet another font change that generates tens of thousands of regression test differences? :-) | 14:40.56 |
henrys | I'm wondering if we need a "pre staff" meeting to filter the list. Seems like a lot to cover especially with new folks. | 14:41.15 |
| marcosw: possibly yes. I was thinking we'd split the difference on the metrics ;-) | 14:41.55 |
| marcosw: I did have a question. Hin Tak got me thinking about using Xvfb for regression testing X11 - as things are now I'm hesitant to do anything to that device. | 14:43.21 |
Robin_Watts | henrys: splitting the difference on the metrics sounds bad. | 14:44.28 |
| Then we'd be wrong on both. | 14:44.32 |
henrys | Robin_Watts: it was a joke. | 14:44.38 |
Robin_Watts | Oh, sorry. | 14:44.47 |
henrys | a metrics database for PCL might be possible. | 14:45.27 |
chrisl | Seems like that would lose a lot of the benefits of merging the font sets...... | 14:46.22 |
Robin_Watts | chrisl: metrics are tiny compared to outlines. | 14:47.02 |
marcosw | henrys: shouldn't be too difficult. Will have to do a bit of work to correctly test multi-page postscript files. luckily we have Ken's code for that :-) | 14:47.03 |
chrisl | Robin_Watts: it means we have something else to maintain | 14:47.20 |
kens2 | My code works on the cluster :-) | 14:47.22 |
henrys | can everyone just use gsview for a couple weeks or so as a regular PDF viewer? I think we'd shake out a few more things if we all just did that. | 14:48.18 |
jogux | henrys: I tried. I emailed Fred some feedback and went back to preview for now... | 14:48.45 |
Robin_Watts | I have been doing that. I've not had to go back to Acrobat. | 14:48.55 |
kens2 | I often have to use Acrobat. Anythign with a transfer function is incorrect, and a fair few recent bugs have involved colour management. | 14:49.45 |
henrys | I didn't have anything else, as the in person meeting draws near review the workflowy. Anybody have more? | 14:50.46 |
mvrhel | hmm Hin-Tak opened two bugs related to the same files. One for mupdf and one for gsview. Robin_Watts do you know anything about 695924? | 14:51.02 |
henrys | marcosw: I'll add xvfb to the workflowy | 14:51.05 |
mvrhel | I think I will close 695962 as a duplicaste | 14:51.40 |
Robin_Watts | mvrhel: That's an epub bug. | 14:52.11 |
mvrhel | right | 14:52.15 |
Robin_Watts | So, one for tor, not for me :) | 14:52.21 |
chrisl | is strongly inclined to ignore anything Hin-Tak says...... | 14:52.23 |
kens2 | chrisl did you reclose the bug from this morning ? | 14:52.37 |
mvrhel | chrisl. so ignore the rant about the files on linux? | 14:52.42 |
chrisl | Not yet | 14:52.43 |
kens2 | I would | 14:52.47 |
chrisl | mvrhel: which files on Linux? | 14:53.06 |
mvrhel | in 695962 | 14:53.17 |
henrys | chrisl: when we get into 136 fonts with the glyph densities of the latest fonts we've received (cyrillic and greek etc.) that is a considerable waste of space. | 14:53.19 |
kens2 | The bundlign of lots of system libs | 14:53.22 |
| We do that deliberately because otherwise you have to build from source or go an install lots of (specific) packages | 14:53.42 |
| mvrhel : You 'could' put that in the bug report if you like, but as Chris says, I'm inclined to just close anythgin Hin-Tak gets involved in myself | 14:54.09 |
mvrhel | right. I am just going to close this and mark as a duplicate for the other mupdf rendering bug | 14:54.25 |
chrisl | We discussed that at length - I pointed out it would not be popular with Linux users, but including all the libs was the decision, so..... | 14:54.36 |
kens2 | nods | 14:54.48 |
Robin_Watts | tor8: The user style sheets stuff could do with some comments about ownership of the buffers involved. | 14:54.55 |
henrys | let's just evaluate hin-tak's bugs like anyone else. He does contribute a bit. | 14:55.22 |
kens2 | He contributes a heck of a lot of noise on the PostScript end of thngs | 14:55.47 |
chrisl | henrys: recently he's been using up more time than he's been saving | 14:55.49 |
henrys | chrisl: you can simply ignore him. | 14:56.09 |
mvrhel | If he is beating on gsview that is fine with me | 14:56.20 |
| I will use him as a low cost tester... | 14:56.32 |
kens2 | Eg ths morning when he reopened a bug, then decided he couldn't debug it and re-assigned it back to us, when Chris had already said he couldn't reproduce it | 14:56.34 |
chrisl | henrys: not when he posts crap - especially on customer bugs | 14:56.47 |
mvrhel | yeah he should stay away from customer bugs probably... | 14:57.07 |
henrys | can we restrict customer edits to staff? | 14:57.37 |
kens2 | Only by making them private I thnk | 14:57.51 |
henrys | maybe marcosw knows | 14:57.52 |
| I'll make a note of it and we'll look into it. | 14:58.26 |
chrisl | He has a tendency to muddy the waters which can be a big pain when trying to work through something with a reporter | 14:58.32 |
henrys | off to the skype meeting. | 14:59.15 |
marcosw | henrys: I believe so. let me check. | 14:59.17 |
mvrhel | bbiab | 15:01.20 |
| in about 1 hour or so actually.... | 15:01.30 |
marcosw | henrys: bugzilla doesn't have such a feature, but I think it would be possible to hack something that would do so. otoh, it means customers can't comment on their own bugs (unless we explicitly allowed the bug reporter to comment on their own bug). | 15:02.31 |
Robin_Watts | Making a bug private means that only gs-priv and people that are directly CCd can comment. | 15:03.06 |
chrisl | We probably want to keep it that way - we occasionally have comments we don't want a customer to see | 15:04.03 |
Robin_Watts | tor8: strdupping the css is bad. | 15:08.54 |
| We should use ref counts like with fz_keep_glyph_cache | 15:09.12 |
tor8 | Robin_Watts: you mean when cloning the context? | 15:11.06 |
Robin_Watts | yes. | 15:11.12 |
tor8 | so the user css is not per context then, but shared? | 15:11.52 |
Robin_Watts | indeed. should be a single thing, IMHO. | 15:14.04 |
tor8 | okay, will fix. | 15:14.34 |
Robin_Watts | tor8: If we use multiple contexts with a single doc (like would happen with multi-threaded rendering), we want the same css at all times :) | 15:16.54 |
tor8 | so, protect the style context with FZ_LOCK_ALLOC as well? | 15:17.25 |
| or not at all; fz_colorspace_context's set functions are not protected | 15:17.59 |
Robin_Watts | tor8: That sounds like an oversight. | 15:18.17 |
tor8 | Robin_Watts: updated commits on tor/master | 15:23.04 |
Robin_Watts | tor8: OK, looks good to me, but don't push the last one. | 15:29.30 |
| It'll break the cluster. | 15:29.42 |
tor8 | Robin_Watts: the last one is for bug 695977 | 15:35.23 |
Robin_Watts | tor8: Yeah, I know. | 15:35.40 |
| I'm inclined to tell him to go jump. | 15:35.47 |
tor8 | it's either that or fix the dependencies somehow, and I've tried and thrown my hands up in disgust | 15:35.52 |
Robin_Watts | tor8: We could always have both mutool draw and mudraw ? | 15:36.04 |
tor8 | mudraw works as a symlink, we could make the 'make install' script create a mudraw symlink | 15:36.20 |
rayjj | does anyone know what 'gs4txtwrite' is doing on casper ? It appears to have been started May 7 and it's using 98+% of the cpu and has a time of almost 17000 sec | 15:36.30 |
Robin_Watts | tor8: I care not for how it works on unix :) | 15:36.38 |
rayjj | it appears to be invoked by the web (apache) and it's user is www-data | 15:37.10 |
| marcosw: that is probably something you know about ^^^ | 15:38.22 |
chrisl | rayjj: I'd have to guess it's related to: http://ghostscript.com/txtwrite | 15:40.21 |
Robin_Watts | tor8: For historical reasons, the cluster invokes mudraw as pdfdraw. | 15:50.12 |
jogux | henrys: should gsview problems be emailed to fred, entered in bugzilla, or something else? | 15:51.52 |
kens | thinks entered in Bugzilla | 15:52.17 |
| maybe assigned to fred there | 15:52.30 |
henrys | jogux: what ken said | 15:52.35 |
rayjj | chrisl: I suspect that chewing up 100% of the CPU is not quite what is intended | 15:53.09 |
henrys | I guess mvrhel is the default assignee for gsview which is fine too. | 15:53.14 |
chrisl | rayjj: probably not, but I know nothing about it other than what I said above..... | 15:53.56 |
jogux | ken: henrys: thanks. entered a bug, then reassigned to fred (as it's OS X) | 15:55.19 |
kens | Apologies for the noise, the thunderstorms seem to be affectng my network connection | 15:57.17 |
Robin_Watts | rayjj: if ( QueryPerformanceFrequency(&freq) == NULL) { gp_get_realtime(pdt); } | 16:08.36 |
| QueryPerformanceFrequency(&freq) returns a bool, so that seems a bit odd. | 16:08.59 |
| There are various warnings from that bit of code as well. | 16:09.10 |
| and from gp_seekable that I'm sure weren't there a while ago. | 16:12.28 |
| actually, that QueryPerformanceFrequency code is just horribly broken. | 16:14.00 |
rayjj | Robin_Watts: that was from an example bit of code. The reference page said: If the installed hardware supports a high-resolution performance counter, the return value is nonzero. | 16:14.04 |
Robin_Watts | right, but zero != NULL | 16:14.25 |
| freq should be a LARGEINTEGER, not a uint64_t | 16:14.46 |
| that's a struct (or a union), not a simple type. | 16:15.12 |
| You *can* probably get away with what you've got, but you need a cast to make it a) stop complaining, and b) indicate that you are meaning to do this slightly dirty thing. | 16:15.51 |
| Also, if the call to QueryPerformanceFrequency fails, you go ahead and call it anyway. | 16:16.47 |
| I think you're missing a 'return'. | 16:17.05 |
rayjj | Robin_Watts: OK, I'll fix it. | 16:17.15 |
| I was sort of in the middle of working on a performance issue for cust 532, so I wasn't paying very careful attention | 16:18.58 |
Robin_Watts | rayjj: Sure. | 16:19.24 |
| Also... while I'm at it... | 16:19.31 |
| in cmd_drawing_color_usage there is a warning there. | 16:19.40 |
| You're casting to (int *) and it should be a (gx_color_index *) | 16:19.53 |
| actually, why cast at all? | 16:20.42 |
| marcosw: ping | 16:25.47 |
| Is fred set up so he can ssh into ATS? | 16:26.28 |
| marcosw: Never mind. I've sorted it. thanks. | 16:42.26 |
marcosw | henrys: it's reasonable straightforward to add custom user rules to bugzilla. should I limit commenting on customer bugs to artifex staff? | 16:49.17 |
Robin_Watts | marcosw: Can we make it so that hintak cannot comment on a customer bug, but everyone else can? :) | 16:55.26 |
henrys | marcosw: yes to artifex staff or a customer author of the bug right? | 17:02.06 |
Robin_Watts | chrisl: It dawns on me that we are seeing Mark Knopfler this weekend. | 17:10.03 |
| You are welcome to stay over (though I think you said you had to get back) | 17:10.17 |
| Do you want to eat here before we go to the concert? | 17:11.10 |
mvrhel_laptop | looking over my bugs. Does anyone want 695441? I am never going to work on this thing. It is marked as Artifex GSView but that is not correct. Henrys, should I assign to you or Marcos to reassign correctly? | 17:14.05 |
henrys | mvrhel_laptop: close it saying bugs should be reported against the "new view" ;-) | 17:16.11 |
| mvrhel_laptop: let me know if you want me to do it. | 17:20.15 |
chrisl | Robin_Watts: Indeed, I have to get home Saturday night - but thanks for the offer. I think it would be good to grab something to eat before the concert | 17:20.38 |
Robin_Watts | chrisl: Plan to be here early enough to eat then :) | 17:21.30 |
| Will talk to the boss this evening about it. | 17:21.41 |
| Pete is off to see the Glasgow concert tonight., | 17:22.07 |
chrisl | Cool! | 17:22.26 |
| I have nothing else planned for Saturday, so just give me a time you want to leave, and I'll get to your place in good time... | 17:22.32 |
henrys | marcosw: there is an xfa project on the horizon for mupdf. | 17:23.50 |
Robin_Watts | support is on at 7:30. NEC is by the airport. Airport is 50 mins away with no traffic. | 17:24.01 |
henrys | marcosw: it's on the workflowy | 17:24.01 |
chrisl | Robin_Watts: It's not the NEC, is it? | 17:24.51 |
Robin_Watts | LG Arena, formerly NEC | 17:25.20 |
chrisl | Ah, right.... | 17:25.43 |
Robin_Watts | So, figure on leaving here at 6? | 17:26.21 |
chrisl | That'll be fine.... I can have a long sleep in! | 17:26.52 |
Robin_Watts | Unless I can tempt you out shooting... | 17:27.20 |
chrisl | Not my thing, really..... | 17:27.39 |
| Have to go.... | 17:28.38 |
Robin_Watts | fair enough :) | 17:28.39 |
| night. | 17:28.41 |
fredross-perry | ok, I am pretty sure something has happened to my ssl configuration at the Artifex end. I can's sftp, ssh or git as I could yesterday. Can someone investigate please? Thanks. | 18:01.40 |
Robin_Watts | fredross-perry: looking now. | 18:02.06 |
fredross-perry | grazi. | 18:02.15 |
Robin_Watts | fredross-perry: Try now, so I can see if it shows in the logs. | 18:04.43 |
fredross-perry | sorry for being a pita. | 18:04.52 |
Robin_Watts | no problem. | 18:05.09 |
| what IP address are you connecting to? | 18:05.19 |
| ok, when I ssh in, I can see an entry in the log saying it let me in. | 18:06.35 |
fredross-perry | just âghostscript.comâ as always. ping tells me 184.73.189.105 | 18:06.47 |
Robin_Watts | OK, same IP as me. | 18:07.02 |
| Would 91.222.16.9 be you? | 18:07.18 |
| or 52.5.142.133 ? | 18:07.35 |
fredross-perry | checking... | 18:07.39 |
Robin_Watts | They may be random portscanners, actually. | 18:07.57 |
fredross-perry | no I think I am 4.35.67.122 | 18:08.19 |
Robin_Watts | Anyway, if you can try to ssh in, I should see it. Hopefully the log will point me at something interesting. | 18:08.32 |
fredross-perry | ok just did it. connection refused. | 18:09.03 |
Robin_Watts | Do it again? | 18:09.24 |
fredross-perry | done | 18:09.31 |
Robin_Watts | 91.222.16.9 is showing in the logs. | 18:09.58 |
| and it says it's trying for user 'root', | 18:10.05 |
fredross-perry | huh | 18:10.17 |
Robin_Watts | Are you by any chance sshing from a shell where you're root ? | 18:10.25 |
fredross-perry | I am definitely doing âfred@ghostscript.comâ | 18:10.33 |
Robin_Watts | Do it again... | 18:11.06 |
fredross-perry | am am not root. I am âfredross-perryâ done again | 18:11.21 |
Robin_Watts | I see: May 18 13:37:29 casper3 sshd[28265]: Received disconnect from 91.222.16.9: 11: Bye Bye [preauth] | 18:11.36 |
| The root thing may have been just a coincidence. | 18:11.48 |
henrys | fredross-perry: ssh -vvv usually tells you something interesting | 18:12.22 |
fredross-perry | nothing interesting. | 18:12.54 |
Robin_Watts | pastebin? | 18:13.08 |
fredross-perry | meaning? | 18:13.19 |
Robin_Watts | paste what ssh-vvv says into pastebin.com and drop the link here. | 18:13.35 |
| I see an "Accepted publickey for henrys". | 18:14.20 |
fredross-perry | http://pastebin.com/uzA0pvfT | 18:14.35 |
sebras | Robin_Watts: I wonder if telnet ghostscript.com 22 would print "SSH-2.0-OpenSSH_6.6.1p1 Ubuntu-2ubuntu2" for fredross-perry... | 18:16.33 |
| Robin_Watts: if not then it may be a router related problem. | 18:16.52 |
henrys | Robin_Watts: just checking if mine was working - it is | 18:16.56 |
fredross-perry | Trying 184.73.189.105... | 18:17.35 |
| telnet: connect to address 184.73.189.105: Connection refused | 18:17.36 |
| telnet: Unable to connect to remote host | 18:17.37 |
sebras | fredross-perry: you didn't forget port 22 at the end of the telnet command right..? | 18:18.03 |
fredross-perry | did not | 18:18.12 |
| maybe this is just a Starbucks thing | 18:18.33 |
Robin_Watts | oh, possibly. | 18:19.08 |
sebras | fredross-perry: they might filter port 22. | 18:19.11 |
fredross-perry | so, I went into another server to which I have access, and did ping from there. ping is able to tell me âfrom casper3.ghostscript.com (184.73.189.105)â, where as mine just says âfrom 184.73.189.105â | 18:20.50 |
| so letâs no sweat this more until I can get home and try it again. Sorry | 18:21.34 |
Robin_Watts | OK, I just tried (from my mac): telnet ghostscript.com 22 | 18:21.50 |
| and I got the expected id string. | 18:21.55 |
fredross-perry | this is not my regular starbux, btw. | 18:22.26 |
| thanks for all the help | 18:22.38 |
Robin_Watts | no worries. | 18:22.46 |
fredross-perry | ok, going dark for a while. | 18:23.05 |
sebras | Robin_Watts: my webserver doesn't serve https so I opted to have a second ssh-server at port 443. but that might not work for ghostscript.com of course. | 18:26.19 |
Robin_Watts | sebras: Yeah, probably minimising the number of holes through which people can poke our server is good :) | 18:27.06 |
sebras | Robin_Watts: yeah, but usually cafés cannot block port 443 since https is used everywhere, so doing something like that may mean you can actually log in. | 18:27.54 |
| Robin_Watts: despite port 22 being blocked. | 18:28.02 |
Robin_Watts | yeah, I understand the idea. | 18:29.18 |
fredross-perry | OK, move from starbux to the cardealership, works now. sorry for the hassle. | 18:30.58 |
Robin_Watts | fredross-perry: No worries. | 18:31.36 |
marcosw | henrys: bugzilla is now configured to only let the reporter, assignee, or an Artifex Staff member comment on bugs that have a non-empty customer field. Let me know if it's not working as expected. | 19:36.50 |
| henrys: tags can still be edited by anyone, but since they "are personal and can only be viewed and edited by their author" that is intentional. | 19:49.02 |
rayjj | Robin_Watts: could you take a look at the patch for the issues you raised earlier (QueryPerformance..., etc.) ? | 19:55.54 |
mvrhel_laptop | henrys: question for you | 20:00.10 |
| bug 695544 | 20:00.18 |
| this is not a gs issue but an issue for Marti and lcms | 20:00.28 |
| should I point that out to the user and close? | 20:00.37 |
Robin_Watts | rayjj: Sure | 20:00.41 |
| mvrhel_laptop: It sounds like a fix that we should apply and then pass upstream to marti. | 20:01.59 |
| rayjj: That looks perfect. | 20:02.55 |
| some of it is very similar to what I have here :) | 20:03.27 |
mvrhel_laptop | Robin_Watts: ok. | 20:03.38 |
Robin_Watts | I mean, it's a genuine issue, AIUI - if you try to build gs on a powerpc, it'll fail without that line, right? | 20:06.35 |
mvrhel_laptop | Robin_Watts: I am looking at the tests in lcms.h and I don't see anything that would require that header | 20:08.11 |
| I mean lcms2.h | 20:08.18 |
| Does chrisl have a powerpc machine to check this? | 20:08.52 |
Robin_Watts | I think he does. | 20:09.02 |
mvrhel_laptop | ok. I am going to make a comment and pass to him to check | 20:09.11 |
Robin_Watts | oh, well, actually, I'm not sure. | 20:09.29 |
| Intel is LE. | 20:09.44 |
| So I thought PowerPC macs were BE? | 20:10.17 |
| Ah, Power PC is generally BE, but new versions can be swapped. | 20:11.04 |
mvrhel_laptop | bug says "The endianness tests in lcms2.h are in particular needed | 20:11.22 |
| to make it work on PowerPC little endian machines" | 20:11.24 |
Robin_Watts | yeah, so PowerPC LE machines are 'less usual'. | 20:11.41 |
mvrhel_laptop | ok | 20:11.46 |
Robin_Watts | Ah. | 20:12.15 |
| I bet endian.h sets __BYTE_ORDER or __LITTLE_ENDIAN | 20:12.28 |
mvrhel_laptop | ok that would do it | 20:12.48 |
Robin_Watts | but possibly we may not have endian.h on all platforms. | 20:13.05 |
mvrhel_laptop | true | 20:13.18 |
Robin_Watts | possibly we want something like: | 20:13.45 |
mvrhel_laptop | so perhaps this would be a make file / autogen issue | 20:13.50 |
Robin_Watts | #ifdef (__powerpc__)\n#include <endian.h>\n#endif | 20:14.13 |
| yeah. Definitely one to run past chrisl. | 20:14.30 |
mvrhel_laptop | ok. I passed it on | 20:14.41 |
| 695810 looks like it could be ugly | 20:14.58 |
| not sure how I could duplicate that one | 20:15.22 |
| is there a way to build with a more strict alignment? | 20:16.09 |
Robin_Watts | ugh. | 20:18.34 |
mvrhel_laptop | my thoughts exactly | 20:19.01 |
Robin_Watts | Double ugh, cos the example he gives at the end is in my code :) | 20:19.37 |
| BUT... I'm hoping those are red herrings. | 20:20.12 |
| cmsBool is an int. | 20:22.07 |
| int BPC[256]; | 20:22.20 |
| well: cmsBool BPC[256]; | 20:22.31 |
| so quite how BPC[i] can give an alignment error I don't know. | 20:22.46 |
mvrhel_laptop | right | 20:23.01 |
Robin_Watts | mvrhel_laptop: Want to throw that one at me ? | 20:24.33 |
mvrhel_laptop | definitely | 20:24.38 |
| thanks | 20:24.41 |
Robin_Watts | no worries. | 20:24.44 |
Robin_Watts | foods. | 20:25.26 |
rayjj | mvrhel: I re-wrote the threshold generation loop. At least I seem to get reasonable looking values now without all of the 'adjust' problems. | 22:04.06 |
mvrhel_laptop | rayjj: great | 23:53.30 |
| henrys: Jung's latest leaks were from his improper of the mupdfnet library | 23:57.29 |
| Forward 1 day (to 2015/05/20)>>> | |