| <<<Back 1 day (to 2016/11/17) | 20161118 |
sebras | Robin_Watts: i never checked whether [ null 0 0 1 ] is in the pdf itself or this is something that tor8's patch synthesizes (in which case maybe that code ought to be changed) | 01:16.51 |
| I discovered the structure when pretty printing the dest argument in some dest parsing function. | 01:17.53 |
Robin_Watts | sebras: /Dest [ null 0 0 1 ] | 11:48.33 |
tor8 | sebras: Robin_Watts: so, now that pdf_print_obj takes a fz_output how do I debug print a pdf_obj from gdb? | 11:51.52 |
| the subclassing means I can't easily look at the contents of the struct either | 11:52.09 |
Robin_Watts | pdf_print_obj(fz_stderr(ctx), obj) ? | 11:54.07 |
tor8 | yeah, that seems to do it. | 11:54.17 |
| but I need to follow up with a fz_write_byte(ctx, fz_stdout(ctx), '\n') as well | 11:54.26 |
| yeah, they're [ null 0 0 1 ] | 11:54.42 |
| Robin_Watts: in pdf_lookup_page_number ... one other way to solve it is to return -1 when the page cannot be found instead of throwing | 11:55.24 |
Robin_Watts | Yeah. | 11:56.17 |
| I'm sure that we should bale out of displaying the whole file if outline loading fails. | 12:02.54 |
| I also feel we shouldn't bale out of loading the outline just because something goes wrong loading a single entry. | 12:03.39 |
| My patch does not completely solve the latter. | 12:04.32 |
| We could try/catch around the node population code in pdf_load_outline_imp. | 12:04.54 |
tor8 | Robin_Watts: see tor/master | 12:05.22 |
| that's a more surgical approach to allow for pdf outline entries that lack destinations | 12:05.45 |
| or have duff destinations | 12:05.53 |
Robin_Watts | Right, but it doesn't address either of the issues above. | 12:06.52 |
| Having pdf_lookup_page_number return -1 on error seems nice to me. | 12:07.17 |
| ftont ddoor, sorry | 12:07.24 |
tor8 | yeah, I could live with pdf_lookup_page_number silently returning -1 on page not found type errors | 12:07.51 |
| hardening the viewers to cope with errors when loading outlines is also worth doing | 12:08.28 |
malc_ | Robin_Watts: your consistent usage of "bale" made me question whether "bail" is really correct, you being English and all that... | 12:10.33 |
kens | Both are permitted | 12:11.08 |
malc_ | kens: thank you, didn't know that | 12:11.37 |
tor8 | malc_: both variants are permitted, though the more common meaning of bale is as in bale of hay. | 12:11.39 |
malc_ | tor8: tack | 12:11.52 |
Robin_Watts | malc_: bail/bale is a disputed point, I believe. | 12:31.00 |
| In a boat, you "bail out". | 12:31.09 |
| I think of "baling out" of a plane as being like a "bale" of hay. | 12:31.57 |
| but I could be wrong. | 12:32.06 |
| tor8: No, having looked at it more closely, I disapprove of pdf_parse_link_dest returning -1 for error. | 12:37.59 |
| tor8: OK, see robin/master for 3 commits (1 yours) that I feel solve this as nicely as we can reasonably hope to. | 12:46.17 |
kens | malc_ Robin_Watts the Cambridge (that's Cambridge in England) dictionary defines 'bail out' as 'to jump out of an aircraft...' but it also says 'UK ALSO bale out'. So it looks like either spelling is correct to me. The US English definition makes no mention of 'bale out'. | 13:12.24 |
Robin_Watts | My OED spends about 3 pages on bail and bale combined, but only mentions water out of a boat (where it says both is acceptable). | 13:19.31 |
| a 'bail' is a shallow bucket apparently. | 13:19.44 |
| it doesn't mention bale out of a plane at all. | 13:19.57 |
| bale apparently means "evil", or "dire" too, which I did not know. | 13:20.41 |
kens | I did know htat. I guess that's an abridged OED (given the number of volumes the real thing runs to). Still I'm surprised it doesn't mention that meaning. Macmillan says its earliest reference is in 1920 | 13:22.48 |
Robin_Watts | kens: It's a 'compact' OED. | 13:24.15 |
| Which is the full thing in 2 books plus a magnifying glass. | 13:24.34 |
| 1971 edition. At the time I bought it newer ones were much more expensive. | 13:25.16 |
malc_ | tor8: "Harden viewers against failures when loading outlines." ENOPARSE | 13:26.09 |
Robin_Watts | akin to: http://www.ebay.co.uk/itm/THE-COMPACT-EDITION-OF-THE-OXFORD-ENGLISH-DICTIONARY-IN-FANTASTIC-CONDITION-/252634337581?hash=item3ad22e112d:g:ToEAAOSwZVlXoid9 | 13:26.23 |
malc_ | tor8: my bad, misread harden as hidden | 13:27.18 |
| tor8: you guys (i.e. swedes) know how to make knives... just bought a carbon mora... it screams | 13:27.50 |
tor8 | Robin_Watts: kens: webster 1913... best free dictionary I've found, though it is lacking in modern jargon | 13:35.58 |
kens | Between us Stella and I have several dictionaries :-) | 13:38.53 |
| But in this case I was keen to use one that definitely had UK ENglish spelling | 13:40.46 |
Robin_Watts | kens: There is another OED compact on ebay for 5 + 5 postage at the moment, should you be so tempted. | 13:40.57 |
kens | I htink we have enough dictionaries at present thanks :-) | 13:41.13 |
| Although I suppose they are all kind of old now | 13:41.47 |
| OED Online wants a subscritpion, I'm not that keen | 13:42.46 |
tor8 | Robin_Watts: with your commit we can't differentiate between real failures that ought to throw (such as running out of memory or other equally nasty problems) and page references that are invalid | 13:56.11 |
| in one case I'd still want us to fail loading the entire outline, and in the other have blank entries | 13:56.25 |
Robin_Watts | tor8: feel free to amend my commits to something more to your taste. | 15:20.33 |
| I had a more complex solution last night that checked fz_caught(ctx) to check for FZ_ERROR_OUTOFMEMORY (or whatever it is), but I didn't like it. | 15:21.19 |
sebras | Robin_Watts tor8: just got a build error due to curl sent to me privately via email. | 16:33.31 |
Robin_Watts | sebras: On what platform? | 16:34.18 |
sebras | I'll got to FamilyMart satiate my Oreo craving before attacking that issue. | 16:34.54 |
Robin_Watts | sebras: At least you haven't got to the stage of ordering biscuits from amazon prime now. :) | 16:35.44 |
sebras | Linux I think. | 16:35.53 |
| Robin_Watts: tor8: I can reproduce the users exact error message on linux by forgetting to run git submodule update --init | 17:00.43 |
Robin_Watts | sebras: Ah :) | 17:00.58 |
sebras | Robin_Watts: I sent a mail so I'm hoping to get a response indicating success. | 17:01.27 |
| Robin_Watts: need to sleep a bit, once I get a response I'll post it here. | 17:33.19 |
| Forward 1 day (to 2016/11/19)>>> | |