| <<<Back 1 day (to 2012/09/06) | 2012/09/07 |
chrisl | kens: I just replied to comp.lang.postscript on that thread about gs on Win95 - current binary releases won't work on Win95...... | 08:37.07 |
kens | chrisl OK that's reaonable, thanks | 08:45.14 |
| I hadn't got round to trying it yet | 08:45.21 |
| ANd really, Windows 95 ?!! | 08:45.28 |
chrisl | I did point out in my reply that it *very* nearly ten years since MS dropped support for it | 08:46.06 |
kens | :-) | 08:46.18 |
chrisl | It's since we changed to VS2005 for the Windows binaries - MS dropped Win95 support in VS2005 | 08:46.48 |
kens | Yes I just read your reply | 08:46.54 |
| I'd forgotten about that | 08:47.01 |
chrisl | I did't add "good luck finding VS2003 anywhere", though! | 08:47.31 |
kens | Not our problem, really, if he wants to run an ancient OS.... | 08:47.48 |
chrisl | I reckon the only place you'll find VS2003 now is in the MSDN "historical" downloads section | 08:48.33 |
kens | I did point out in my reply that running an archaic OS is fine, as long as you don't want to ever upgrade any of the software, but then, that's exactly what the original poster wanted to do.... | 08:48.33 |
| OK the debugging prints in my hint writing code say I'm writing the hints correctly, lets see what my reader makes of it. | 08:50.18 |
| Nope still wrong | 08:51.20 |
| of course, its as likely to be the reading code as the writer ;-) | 08:51.47 |
chrisl | kens: In theory, evince should parse and use the hint stream (given that poppler/xpdf does so), if you stick the file somewhere I can try it in evince | 08:54.36 |
kens | chrisl let me check it myself a bit first. I'm willing to bet it doesn't parse the whole hint stream though | 08:55.09 |
| After all, it has to cope with the garbage that Acrtobat creates | 08:55.26 |
| Definitely out by 2 bytes when reading ti back | 08:56.57 |
| Should be 00 00 00 04 getting 00 04 00 00 | 08:57.13 |
chrisl | kens: poppler does seem to do a fair amount with the hints stuff: http://cgit.freedesktop.org/poppler/poppler/tree/poppler/Hints.cc | 08:58.32 |
kens | Hmm, I wonder how it likes Acrobat' rubbish files ;-) | 09:00.19 |
| Anyway, I think I see 'a' problem. | 09:00.33 |
| chrisl in /home/ken on casper, a file called gs.pdf would you mind trying that one with Evince please ? | 10:03.10 |
chrisl | kens: evince displays it with no errors/warnings - xpdf crashes! | 10:10.31 |
| I guess the poppler revised code quietly ignores issues with it. I can do a debug build of poppler and try to see what it's doing if you like | 10:11.34 |
kens | chrisl fascinating.... | 10:11.44 |
| I'd be interested, crashing is bad though :-) | 10:11.57 |
| While I'm prepared to believe there are bugs in it, I don't think its *that* bad | 10:12.39 |
chrisl | The xpdf, I suspect, is not very up to date.... | 10:12.47 |
| kens: this may take some time - the poppler build is totally autotooled, and I'm struggling to work out how to build non-optimized...... | 10:37.44 |
kens | chrisl don't worry | 10:38.30 |
tor8 | kens: what time are you getting in @SFO tomorrow? | 10:47.42 |
| I'm arriving early for once, so this time I can hopefully join for dinner | 10:48.20 |
Robin_Watts | ah. online checkin time. | 10:48.21 |
| we land at 13:25 | 10:48.44 |
tor8 | Robin_Watts: that gives us plenty of time to meet up at the hotel then :) my arrival is with air france at 12:45. so let's pray they don't have a strike in paris... | 10:49.47 |
kens | Not Lufthansa at least | 10:50.00 |
tor8 | kens: thankfully, if it were I'd be worried by now | 10:50.11 |
Robin_Watts | checks in with a sneaky seat change to hopefully get a space next to him. | 10:52.31 |
kens | How did you manage a seat change ? Won't let me | 11:05.40 |
Robin_Watts | kens: Had you printed a boarding card? | 11:05.53 |
kens | Not yet | 11:06.02 |
chrisl | It will let you after you check-in, but before you issue the boarding pass | 11:06.05 |
kens | getting there now | 11:06.06 |
Robin_Watts | Hmm. Dunno then. | 11:06.13 |
kens | Not quite checked in yet either | 11:06.24 |
chrisl | kens: I can now debug the poppler code - but I'm not sure what I'm looking for..... | 11:06.42 |
Robin_Watts | I checked in, then changed seat, then printed boarding pass. | 11:06.44 |
kens | chrisl I don't know :-) | 11:06.53 |
| Whatever causes the crash I guess... | 11:07.03 |
chrisl | That was xpdf, I'm looking at poppler - I could get xpdf, I guess | 11:07.30 |
kens | chrisl seriously, don't worry | 11:08.01 |
| I'm going to commit my code shortly and fix any problems later | 11:08.22 |
chrisl | So, for example, from the hint stream, the page offsets are 531 for for th first page, and 790 for the second - does that sound like what you expect? | 11:08.27 |
kens | After all 'Acrobat can open it' :-) | 11:08.33 |
| Ye those numbers are ~correct | 11:08.42 |
chrisl | The shared objects stuff is confusing...... | 11:10.16 |
kens | Its also probably wrong | 11:10.23 |
| THat's what I'm working on at teh moment | 11:10.31 |
chrisl | It seems to think there are 5 "shared groups"?? | 11:11.01 |
kens | None of the available seats look like an improvement, I'm staying in 62G | 11:11.15 |
| chrisl yes, probably | 11:11.23 |
chrisl | That's what I thought.... | 11:11.28 |
kens | all objects on page 1 are considered 'shared' and I only enter one object oper group | 11:11.40 |
chrisl | None of the shared groups have signatures.... | 11:12.41 |
kens | also correct | 11:12.47 |
chrisl | And, yes, each group has only 1 object in it | 11:13.30 |
kens | That's actually pretty good. | 11:13.37 |
| My reading code must be wrong | 11:13.43 |
sebras | Robin_Watts: tor8: what does the plan for the forms stuff look like at the moment? | 11:15.46 |
chrisl | First group offset is 531, second is 66067, then 131603, 197139, 262675 - hmm, questionable..... | 11:16.19 |
sebras | it got merged to origin/master the other day, and I guess this means I ought to look at documenting anything not yet documented, no..? is there a deadline? ASAP? | 11:16.25 |
kens | offsets are totally wrong | 11:18.26 |
| But still, I'm surprised its that good ;-) | 11:18.35 |
chrisl | Given that it gets the page offsets correct, it *may* be reading the shared group offsets correctly, too | 11:19.19 |
tor8 | sebras: don't document the forms stuff yet. | 11:19.31 |
kens | I suspect it is, and I'm writnig them wrong | 11:19.32 |
chrisl | Yeh, still, it does seem we have *something* to compare against.... so does Acrobat now agree that the file is linearized? | 11:20.11 |
kens | Yes it does, Acrobat resally doesn't use the hint streams | 11:22.28 |
chrisl | So what was causing Acrobat to hiccup on your output before? | 11:22.56 |
kens | Didn#'t you see that one ? You'll likeit..... | 11:23.11 |
| Acrobat won't recognise any file < 4Kb as linearised | 11:23.21 |
| Add white space to make the file 4096 bytes and Presto! file is linearised. | 11:23.43 |
chrisl | Oh, you've *got* to be kidding - that really is a new low...... | 11:23.50 |
kens | Two days of hacking and crafting PDF files, no I'm not kidding :-( | 11:24.08 |
chrisl | Just when you think Acrobat's insanity can't surprise you any more | 11:24.33 |
kens | That one had me for quite a while.... | 11:24.53 |
| I couldn't quite bring myself to believe it | 11:25.00 |
chrisl | I'm not surprised - although it does explain why they were less than worried about making a single page file *bigger* in order to be "web optimized" | 11:25.48 |
sebras | tor8: noted. | 11:33.21 |
Robin_Watts | sebras: We hope to ship an alpha mid october. | 12:38.35 |
| tor8: What (in your view) needs to be changed before we start documentation? | 12:38.49 |
kens | Robin_Watts : tor8 how do I squash commits again ? | 12:59.34 |
tor8 | Robin_Watts: I think we should have the gtk+ viewer with proper forms support ready; until then the APIs may well need revising | 13:03.14 |
| Robin_Watts: which reminds me, I ought to focus on the viewer again... ugh. | 13:03.31 |
| kens: git rebase -i | 13:03.38 |
Robin_Watts | tor8: right, but how much revision do you expect it to take? | 13:04.21 |
tor8 | Robin_Watts: no idea, that's why I think we should hold off. my biggest fear is that sumatra may jump on it and start hacking away... | 13:04.54 |
kens | tor8 thanks | 13:10.28 |
| Hmm just gives me a list of options | 13:12.25 |
Robin_Watts | kens: git rebase -i HEAD~10 will give you a list of the last 10 things in an editor. | 13:14.41 |
| You reorder them, and change 'pick' to squash etc, then when you save and exit the editor it does it. | 13:14.59 |
kens | yeah I think that's what IO need | 13:15.10 |
| I forgot the HEAD~n | 13:15.33 |
| Well, its doing 'something' :-) | 13:22.26 |
| Hmm sems to have ended up as 4 commits, I must have made a mistake, still its better than 11 | 13:28.17 |
Robin_Watts | kens: Well, you can always rebase again. | 13:31.18 |
kens | I think they all build and run so I don't mind | 13:34.49 |
Robin_Watts | Morning mvrhel. I have tea. | 13:39.52 |
mvrhel | Robin_Watts: oh great. thank you. | 13:40.15 |
Robin_Watts | np. | 13:40.22 |
sebras | Robin_Watts: tor8: alright, good to know. | 14:01.43 |
kens | Drat one of my commits has ended up with loads of seg faults | 14:02.26 |
| ;-( | 14:02.35 |
henrys | I didn't know gdb supported python scripting now, chrisl do you ever use that? | 15:05.48 |
chrisl | henrys: I tried it when it first appeared, but I never got it work properly - I haven't tried it since | 15:06.33 |
henrys | gdb 7 also has the reverse debugging but it's pretty pokey compared to undo-db | 15:07.12 |
chrisl | I think the reverse debugging I wouldn't expect to be fully useful for some time | 15:08.04 |
henrys | I forgot to say in the agenda - submit edits and additions - hopefully folks just know to do that. | 15:08.41 |
Robin_Watts | Jules and Greg have a patent on their reverse debugging technique. | 15:09.17 |
| so gdb ain't ever going to be as good as undo-db (factor of 1000+ I believe). | 15:09.52 |
henrys | Robin_Watts, yes I remember you saying that. Once in a while I use their product. | 15:09.55 |
kens | Robin_Watts : patents expire :-) | 15:10.31 |
henrys | do you have a pointer to the patent? I didn't see it googling for the obvious. | 15:12.46 |
Robin_Watts | no, I'm afraid not. | 15:14.22 |
| http://www.google.com/patents?id=q98AAgAAEBAJ&printsec=frontcover&dq=Julian+Smith+debug&source=bl&ots=AhQ-m22vvV&sig=IuwSb58FYcV9EqkwldL_eVj0r4U&hl=en&sa=X&ei=oQ9KULmQI-Oe0QXvgYFY&sqi=2&pjf=1&ved=0CC4Q6AEwAA | 15:16.02 |
henrys | ugh hcl is an example of a company we didn't want to ask ;-) | 15:17.44 |
| no new money and more p2 bugs | 15:18.09 |
kens | Yeah. | 15:18.26 |
| But they one they have open at the moment is easy enough, its not fixable. | 15:18.40 |
| THe answer basiclaly is 'don't do it like that' | 15:19.03 |
henrys | they do just plug stuff in bugzilla and don't use support right? | 15:20.26 |
kens | So far, yes, this is the first time we've heard that they are working on behalf of a custoemr | 15:20.45 |
| I'd be inclined to check with the cuystomer, but I guess its probably legitimate | 15:21.13 |
henrys | yes I responded with "if this is all correct ..." but I'll leave the rest to Scott and Miles | 15:23.10 |
ray_laptop | henrys: Our contract calls for a single point of contact for customer bugs. HCL should have the customer submit bugs rather than just claim they are working for a customer. Who knows, it may be a different group or product to the supported customer even if it's in the same company | 15:30.14 |
| note that we have two separate companies with that same parent corporation -- cust 534 and 535 so at the least we need to know which, and it's better for us if we are in contact with our customer rather than through some intermediary like HCL | 15:33.00 |
kens | Especailly since it took HCL 3 attempts to even report a bug which made sense | 15:33.23 |
| :-) | 15:33.25 |
henrys | it looks to me like the original customer has outsourced the project given geography and other clues so the contract would be good, but I just spoke to Scott about it and told him to tell Miles to sort it out. | 15:35.44 |
| ray_laptop, kens:I think Miles understands the issues and he's always eager to get more money so I imagine he'll look at all the angles, so I'm pretty sure this will get handled okay. | 15:38.17 |
ray_laptop | BTW, I get into SFO tomorrow at 3:45, so let me know if a group is going to dinner. | 15:38.49 |
| henrys: right. I'd just rather be in contact with our customer rather than some software consultancy that has a contract with our customer | 15:39.54 |
kens | ray_laptop : we're in at 1`3:25 and tor even earlier, so we were planning to go out for dinner | 15:40.16 |
henrys | I don't arrive until 7:40 so I'll probably see ya'll in the morning. | 15:40.54 |
| funny they just use our pdfwriter as a compiled exe - they have to outsource that? | 15:42.30 |
kens | OK good all seg faults gone :-) | 15:49.36 |
Robin_Watts | tor8: ping | 15:54.17 |
| In fz_device terms, is a mask operation matched with a pop_clip ? | 15:54.41 |
ray_laptop | henrys: I didn't think either cust 534 or 535 groups was using our pdfwrite -- at least one of the groups uses us to convert PDF to PS2 (ps2write) | 15:55.17 |
Robin_Watts | Ah. mask, end_mask then pop clip. | 15:55.36 |
henrys | ray_laptop:yes I think you are right - but as I recall they don't even build the code just use it as is. | 15:56.41 |
tor8 | Robin_Watts: yes. | 15:57.07 |
Robin_Watts | That explains it. | 15:58.09 |
| Should we be passing a colorspace into a begin_group ? | 15:58.30 |
tor8 | Robin_Watts: for the blend mode colorspace? probably. | 16:25.24 |
Robin_Watts | mupdfwrite does transparency groups now. | 16:26.40 |
kens | cool | 16:27.08 |
FreezingCold | How many pixels is 612 x 792 pts? | 16:38.24 |
kens | depdensd on resolution | 16:38.44 |
| OK I'm off, see you all tomorrow or Sunday | 16:43.00 |
mvrhel | hehe I like the google star trek sounds | 17:10.57 |
| Forward 1 day (to 2012/09/08)>>> | |