| <<<Back 1 day (to 2016/09/10) | 20160911 |
Hello71 | huh? | 00:10.58 |
| to be painfully clear, I mean that from a *user* perspective, it is easiest just to xz the whole file | 00:11.33 |
ray_laptop | Hello71: if all you are looking to do is compress files, fine, but PDF readers will *not* automatically decompress an .xz file | 00:12.57 |
Hello71 | yes, of course | 00:13.26 |
ray_laptop | Hello71: well "of course" for you wasn't clear. Most people want to create portable PDF files (and possibly with minimal size or specific compression / downsampling ) | 00:14.30 |
Hello71 | but I can write a script to do xz -dc thing.pdf.xz > /tmp/thing.pdf; mupdf /tmp/thing.pdf | 00:16.52 |
ray_laptop | Hello71: seems that you are at a much lower level. Of course you can script that, and if you want to do a little programming you can modify mupdf so that it can take in stdin and pipe to your new mupdf and let the OS create a safe tempfile | 00:19.55 |
Hello71 | afaict mupdf needs a real file | 00:21.22 |
ray_laptop | Hello71: or modify mupdf so it uses "magic" header type detection to detect .xz files and decompress to a safe tempfile | 00:21.23 |
Hello71 | too much work, bash script is easier | 00:21.45 |
ray_laptop | Hello71: lazy. Error prone except for "quick and very dirty" single user mode. hard coding /tmp/thing.pdf is *not* sage | 00:22.36 |
| s/sage/safe/ | 00:22.45 |
Hello71 | well, it's perfectly *secure* | 00:23.35 |
ray_laptop | not if you try and have two shells open running the same script | 00:24.12 |
| I tend to think in more general solution terms | 00:24.40 |
Hello71 | yes, either both will work, one will work and one give error (or segfault, or give mupdf error), or both will error | 00:25.09 |
| but you cannot compromise security, it just won't work right in some cases | 00:25.28 |
ray_laptop | runs screaming | 00:25.34 |
Hello71 | did I tell you, I also write financial webapps that only work in one tab | 00:26.10 |
| if you try to ctrl click it logs you out | 00:26.29 |
ray_laptop | I am not into user abuse | 00:27.23 |
pluc_pilou | Thanks ray... Not patient enough. I'll look into that. | 01:03.06 |
sebras | ray_laptop: still around? | 01:35.45 |
| tor8: hm... changing pdf_set_markup_annot_quadpoints() and pdf_set_ink_annot_list() to use float* instead of fz_point makes it much easier. | 07:42.31 |
| but I'm pondering whether it would make more sense to use fz_path (and consequently Path in java). | 07:42.49 |
| maybe not as linecaps and joins are not handled in the same way. well, it would have been a nice interface to use at least. :) | 07:46.26 |
| tor8 (for the logs): ugh... after having implemented quad points get/set for markup annots I noticed that you have started a new interface. :-/ | 09:17.40 |
| tor8: however pdf_annot_vertex_count() and pdf_annot_vertex() is not enough for InkAnnotations as they may have several _disjoint_ paths. | 09:18.56 |
| I think we both need to be online an discuss this a bit. | 09:19.16 |
| I have a number of commits on sebras/wip based on your tor/master commits (which were rebased on top of origin/master). | 09:20.06 |
| where I add date/author/borderstyle things and attempt to implement these in JNI too. | 09:20.29 |
deekej | OT: you working even on Sunday, guys? :) | 09:23.34 |
sebras | deekej: most of us are not, but I am. | 09:23.57 |
deekej | sebras: ouch... but yeah, I'm working as well right now, catching up :) | 09:24.30 |
sebras | deekej: for me it is more like catching ahead I think. :) | 09:24.56 |
| Forward 1 day (to 2016/09/12)>>> | |