| <<<Back 1 day (to 2013/12/10) | 2013/12/11 |
sebras-mobile | Good morning. | 08:48.03 |
| kens: morning! | 08:54.34 |
kens | Hi sebras | 08:54.41 |
sebras-mobile | kens: have you sent tor back to .se now? | 08:55.09 |
kens | Dispatched him yesterday | 08:55.24 |
| He should have been home yesterday evening | 08:55.34 |
| I expect he's even more tired than me though, it may be some time before he surfaces...... | 08:56.00 |
sebras-mobile | kens: yey, then I can bother him about HW soon. (I asked him to buy me a spare e-ink devuce) | 08:56.17 |
kens | sebras, we couldn't find anywhere stocking the new Kobo, we looked in all the airports | 08:56.42 |
sebras-mobile | kens: no worries, I'm waiting for the breakfast break in my SCRUM retrospective course so I can join it... | 08:57.27 |
kens | SCRUM ? | 08:57.42 |
sebras-mobile | kens: that's alright. | 08:57.45 |
| Yeah, companies in .se insists on doing this... :/ | 08:58.06 |
kens | Oh, its a development thing | 08:58.29 |
| chrisl ping | 10:28.37 |
chrisl | kens: pong | 10:29.31 |
kens | Did you suggest that I should hand #694836 to you ? | 10:29.48 |
chrisl | Is that the one with the two fonts with same name? | 10:30.21 |
kens | Yes, that's the one, even Acrobat complains when opening the file | 10:30.37 |
| Ah one is a CIDFont, and one is a font | 10:31.01 |
| Neityher is embedded | 10:31.06 |
chrisl | Nice, small file :-( | 10:32.27 |
kens | Yeah, this customer always sends these kinds of things | 10:32.41 |
| I'm not convinced this is a font problem | 10:33.36 |
| The text is missing because it throws an error in the content stream | 10:33.48 |
| And so aborts the stream | 10:33.57 |
| THere is a lot more than just text missing | 10:34.37 |
chrisl | Evince gives all kinds of errors, too. | 10:34.56 |
kens | Just debugging this looks like its going to be awkward, I bet if I edit it in Acrobat the problem goes away | 10:35.35 |
| and huge amounts of the content are drawings of scaled 1x1 strokes | 10:36.16 |
| Oh they aren't even scaled, just use cm to translate the position. | 10:36.37 |
| What an ugly file | 10:37.12 |
| OK I htink I answered my question, I'll keep this one. At least until I can prove its a font problem | 10:38.14 |
chrisl | It's using "/Registry (PDFAUTOCAD) /Ordering (Indentity0)" which may be causing us problems..... | 10:39.20 |
kens | COUld be, interestingly if I just run page 2 it works | 10:39.45 |
| WHich does point towards fonts | 10:40.12 |
chrisl | My computer is horribly slow for some reason..... really thrashing the hdd | 10:46.11 |
kens | I started running the file with -dPDFDEBUG and -dPDFSTOPONERROR, its still going | 10:49.06 |
| Not finished page 1 yet | 10:49.19 |
| ah page 1 done | 10:49.41 |
| Seems some climate scientist and elf-botherer has been having some fun: | 10:52.05 |
| http://www.theregister.co.uk/2013/12/11/los_angeles_weather_like_mordor/ | 10:52.05 |
| I wonder how Ray likes living in Mordor | 10:52.17 |
chrisl | I saw that the other day | 10:52.38 |
kens | OK the error is a font error | 10:54.09 |
| "Can't build /Identity0.Unicode /CIDDecoding resource" followed by an undefinedresource in findresource | 10:54.54 |
| THis is caused by the CIDFont "Arial-ItalicMT" object 19 | 10:55.36 |
chrisl | Well, what is "Identity0"? | 10:56.11 |
kens | Make that 'Indentity0 :-) | 10:56.36 |
chrisl | Yeh, actually, I suspect Identity0 might work | 10:56.51 |
kens | I'm not sure why this doesn't fail if I only run page 2 though | 10:57.10 |
chrisl | If I replace Indentity0 with Identity0 the file works fine | 10:58.44 |
kens | Taslk about a broken file | 10:59.38 |
| So why does it work if I only run page 2.... | 11:00.01 |
chrisl | No idea | 11:01.04 |
| TBH I can't remember why Identity0 works...... | 11:03.34 |
tor5 | kens: speaking of mordor. http://kieranhealy.org/blog/archives/2011/07/29/text-editors-in-the-lord-of-the-rings/ | 11:03.37 |
| kens: got home safely yesterday, to end up in train chaos... 3/4 of trains cancelled :( | 11:03.57 |
kens | tor5 really sorry to hear that :-( | 11:09.51 |
Pulse__ | Is there an easy way to add a stamp (a text string) to every page in a multipage pdf (or ps, if that is easier) using GPL Ghostscript 9? | 11:11.12 |
kens | Use an EndPage procedure | 11:11.32 |
| Works for both PS and PDF input | 11:11.44 |
Pulse__ | using the -c flag, or do I need to create a script for this? | 11:14.15 |
kens | You can use -c to inject PostScript, or you can write it the PostScript a separate file and insert it before your input file, either should work | 11:15.03 |
Pulse__ | ok, thanks! I will try to play around and see if I can figure out how to do it. | 11:16.50 |
kens | chrisl the working version falls bacl to an Identity ordering, the non-working one doesn't. It looks like the difference is something to do with having already created PDFAUTOCAD-Indentity0 | 11:19.13 |
chrisl | kens: yes, but I'm confused about what's going on there | 11:23.13 |
kens | chrisl it 'looks like' we already defined the font once on page 1, so we try to use it on page 2 and it fails, not sure why | 11:23.45 |
chrisl | kens: that's what I mean, I can't see why using the font again should cause an error - I could see the output being wrong, but.... | 11:24.45 |
kens | yeah its the fact that it tries to substitute agin, I'm baffled by it. However it *does* look font related, can I assign it to you ? | 11:25.21 |
chrisl | Sure | 11:25.32 |
kens | OK thanks. | 11:25.39 |
| chris, I need to build the commercial version of GS in order to get the Luratech decoder, which I haven't done for absolutely ages. Any clues ? (eg where do I get the relevant source...) | 11:59.54 |
chrisl | The luratech code is in subversion, you put it into the gs directory | 12:02.06 |
kens | Oh, guess I need to install subversion then.... | 12:02.23 |
chrisl | I can e-mail it, it's not huge | 12:02.35 |
kens | THat would be quicker probably, thanks | 12:02.47 |
chrisl | Hmm, actually, it's rather bigger than I thought..... | 12:06.22 |
kens | OK then I'll go get subversion | 12:06.31 |
chrisl | No, it's on it's way - it was just big enough to trigger the gmail "security" thing, so had to rename the file | 12:07.41 |
kens | ah, thakns | 12:07.49 |
| yep trickling in now | 12:07.58 |
chrisl | So, unzip it into the ghostpdl/gs directory, and then do a clean rebuild, and it should build in automatically | 12:08.18 |
kens | Even under Windows ? O.O | 12:08.36 |
chrisl | Yes | 12:08.51 |
kens | OK building now | 12:09.44 |
chrisl | nmake allows us to use conditionals and also to check for files/directories existence | 12:10.03 |
| So we can do: !if exist("luratech\ldf_jb2")..... | 12:10.23 |
kens | :-) | 12:10.35 |
chrisl | Famous last words, of course! | 12:11.02 |
kens | I'll let you know.... | 12:11.18 |
| Well it says it built.... | 12:13.12 |
| Seems to w3ork too, thanks chrisl | 12:16.08 |
kens | lunches | 12:16.24 |
chrisl | NP, I'm going out for a little while, too | 12:16.37 |
Robin_Watts | morning all | 13:20.14 |
kens | welcome back robin | 13:20.22 |
Robin_Watts | How was your trip home ? | 13:20.45 |
kens | unpleasant | 13:20.54 |
Robin_Watts | any particular reason? | 13:21.15 |
kens | many :-) | 13:21.21 |
Robin_Watts | They downsized our first airplane because of the snow etc. 47 fewer seats. | 13:22.06 |
| So our row disappeared, which they didn't tell us til we were about to board. | 13:22.27 |
| They sorted it all out, and we were fine, but the plane left an hour late, and that hour was full of obnoxious shouting americans complaining that their first class seats were changing. | 13:23.05 |
kens | we had that problem in LAX, fortunately Tor had asked the crew about presenting passports, so we were in the front of the chaos | 13:23.33 |
| Also luckily the crew on the aircraft were able to tell us about an airside shuttle between termnals, so we didn't have to go through airport security again at JFK we would probably have missed the flight if we had to. | 13:24.42 |
| The inflight entertainment on the JFK->LHR stuttered all the time | 13:25.25 |
| and the 'entertainment' on LAX->OGG->LAX was the 'shared overhead screen' style, which is just awful. | 13:26.04 |
Robin_Watts | The united flights were excellent. I am seriously considering never flying virgin again. | 13:26.09 |
kens | Would mean changing in the US all the time. Not interested in that | 13:27.11 |
Robin_Watts | United 931: direct from LHR to SFO | 13:27.44 |
tor5 | Robin_Watts: so having to suffer US airlines incessant self promotion and bragging about "core values" and messages from their CEO doesn't annoy you? ;) | 13:33.31 |
Robin_Watts | tor5: No more than beardy branson does. | 13:35.38 |
| Seatback power trumps everything, pretty much. | 13:35.56 |
tor5 | Robin_Watts: ah yes, but you only see beardy branson if you try to watch an in-flight movie, not at inescable lound volumes while waiting for the flight to get off the ground... | 13:36.23 |
| Robin_Watts: get a better battery life laptop ;) the chromebook c720 reportedly has 8.5 hours. | 13:37.21 |
| Robin_Watts: did you get back today? | 13:38.01 |
Robin_Watts | tor5: but there are no games for that worth playing, and even if there were, the battery life would evaporate. | 13:38.06 |
| tor5: We made it back at midnight last night. | 13:38.18 |
| have just surfaced. | 13:38.23 |
tor5 | Robin_Watts: steam for linux works with crouton | 13:38.24 |
Robin_Watts | The world is gently spinning around my head at the moment. | 13:38.35 |
tor5 | but yeah, battery life will evaporate rather quickly | 13:38.36 |
Robin_Watts | so, I did some work on the JNI stuff on the way home, and hit a problem. | 13:40.38 |
| The pickling of the context into the document/device etc. | 13:41.38 |
| Either we need to require java callers to follow the same 3 rules of mupdf, or we need to take steps to solve it. | 13:42.21 |
tor5 | TLS? | 13:42.32 |
| or is that not enough? | 13:42.37 |
| TLS as in hiding the details of having a context altogether from the Java interface | 13:42.55 |
Robin_Watts | We can solve the pickling of the context into the document/device with new calls to rebind the document/context and device/context. | 13:43.06 |
| Indeed, that lets us hide the context entirely. | 13:43.18 |
| With the PDF level interface, though, I can't see how to make that work. | 13:43.30 |
| because in java you'd say: "get me blah from this pdf_object", and that'd hit an indirect object, and that would need a context, and would get the document one. | 13:44.39 |
| We have no 'document' at the PDF level to be able to rebind at. | 13:44.55 |
| (Am I making any sense?) | 13:45.01 |
tor5 | Robin_Watts: yeah. we'd need to assert() or similar that the pickled context is the same as the thread local one | 13:51.16 |
Robin_Watts | I was handling the thread local stuff nicely in the java level. | 13:52.33 |
| which means to do the 'assert' thing, I'd need to have new C level entry points to pass the java level context into the C. | 13:53.01 |
| Which leads me to the idea of having a new set of 'pdf_object' functions that take a context. | 13:53.28 |
| If we're going for that, then they can pass the context around internally and rebind if required. | 13:54.22 |
tor5 | Robin_Watts: hmm. there are several other things than just the exception stack in the context, I'm not sure rebinding is a good idea at all. | 14:17.03 |
| a pdf_object's context is tied to the same context that resolve_indirect needs, rebinding it would be bad | 14:17.24 |
Robin_Watts | tor5: The only thing in the context that isn't shared between threads is the exception stack. | 15:19.40 |
| Hmm. Maybe there is a smarter way to work here. | 15:20.31 |
| I can vaguely see the outline of an idea, but I don't entirely trust myself in this jetlagged state. | 15:20.59 |
| tor5: Is that a new mobliey number? | 16:47.32 |
ray_laptop | marcosw: I replied to the email -- I'll look for the product key as soon as I get in to the office (oil change in process) | 17:05.33 |
Robin_Watts | ray_laptop: AIUI, the bug that underlay the need for that has gone away now, right? | 17:06.11 |
ray_laptop | Robin_Watts: it went away for cust 32, but cust 1 still has the issue (unless they also are doing something wrong). It's curious that cust 32 didn't mention calling gsapi_set_arg_encoding, but it was a *VERY* sketchy email report | 17:08.44 |
| Robin_Watts: but thanks for the clarification in the bug | 17:09.10 |
Robin_Watts | no worries. | 17:09.23 |
ray_laptop | From the bug report, it seems that cust 1 is able to get the error using a command line from a shell | 17:10.30 |
Robin_Watts | ray_laptop: Are they using a WINDOWS_NO_UNICODE build? | 17:13.12 |
ray_laptop | Robin_Watts: I don't know. They said they are using 9.10. | 17:16.09 |
kens | chrisl nice fix | 17:16.12 |
chrisl | kens: it's what I should have done in the first place, but, hey, it's done now..... | 17:16.35 |
ray_laptop | hmm... I just had a look at the ps.log attachment. They are using -dUseCIEColor with pdfwrite, which ISTR is not recommended. Checking the tracker now... | 17:17.28 |
chrisl | Hrm, that log shows them using ps2pdf - I wonder if that is relevant..... | 17:19.08 |
| kens: when you tested with the relevant directory name did you call ps2pdf or call gs directly? | 17:20.44 |
kens | directly | 17:21.33 |
| ray_laptop : its not recommended with *current* code | 17:22.00 |
ray_laptop | kens: thanks. So with 9.10 it was OK ? | 17:22.23 |
henrys | ray_laptop: was another parameter intended here or should the last specifier be removed?: if_debug3m('l', cldev->memory, "[l]writing for bands (%d,%d) at %ld K %d \n", band_min, band_max, (long)cb.pos); | 17:22.28 |
kens | should be yes | 17:22.29 |
| OK heading off, bye all | 17:29.18 |
ray_laptop | bye, kens | 17:30.38 |
| henrys: That looks strange | 17:30.48 |
| I see what I did -- it used to also print out the 'band_compliexity.uses_color. I got rid of that, changed to debug3m (from 4m), but forgot to edit the format string. | 17:37.52 |
| henrys: how'd you spot that ? | 17:38.29 |
henrys | I was looking -ZL and saw it was uninit | 17:38.58 |
ray_laptop | henrys: OK. Thanks. I'll patch that, unless you want to. | 17:39.35 |
henrys | ray_laptop: nope go ahead | 17:40.06 |
Robin_Watts | henrys: Bugs 694124 and 694363 - shellys fixes look good to me. | 17:57.17 |
| I have a minor query on 694119 | 17:57.28 |
| They are assigned to you, but I would be happy to take on getting the fixes committed etc. | 17:57.59 |
| Aha! All pdf_object's have a pdf_document now. | 18:42.34 |
| That makes life MUCH easier. | 18:42.41 |
| oops. Bad use of '. I blame the jetlag. | 18:43.50 |
henrys | Robin_Watts: sure go ahead | 18:55.58 |
| thanks | 18:56.00 |
Robin_Watts | When tor5 surfaces, I have an idea about a nice way to do the locking required for the JNI bindings etc. I think. | 19:01.32 |
| but I suspect he's at least as lagged as me. | 19:02.00 |
ray_laptop | Robin_Watts: I just sent a patch off to cust 801 that (I think) fixes the 'raw' output. It was _seriously_ broken since it didn't allow for the *in buffers being padded. | 20:55.38 |
| Robin_Watts: you didn't write anything to convert their 'raw' files did you ? (If not, I think I'll do one -- at least for the CMYK planes) | 20:56.52 |
mvrhel_laptop | ahh found a HUGE leak in my printing with the windows 8 app. I was wondering why the surface was always crashing... | 21:55.46 |
| Forward 1 day (to 2013/12/12)>>> | |