| <<<Back 1 day (to 2015/09/02) | 20150903 |
chrisl | kens: I am going to remove pdf_cslayer.ps - given that it's never been in our builds, and (as Ray pointed out) given everything else we've changed this release, this is a) easily reversible and b) least likely to cause issues for any current users | 07:58.40 |
kens | Sounds good to me | 07:59.12 |
| I'm unconvinced it would work with the PDF interpreter any more anyway | 07:59.29 |
chrisl | It "worked" (inasmuch as it didn't cause any interpretation errors) last year - but it had no useful effect, as far as I could tell | 08:00.43 |
kens | I thnk it might cause problems because it doesn't track the BMCLevel, which the regular PDF interpreter implementations do | 08:01.13 |
| I have no idea if that would really cause a problem, but I'd be concerned that it might, when processing certai PDF files | 08:01.48 |
chrisl | As no device actually handles what it sends, I'm thinking no effect at all | 08:02.37 |
kens | On the device, no, but I suspect other parts of the PDF interpreter use the BMCLevel, which would not be set by the pdf_cslayer implementatons of BMC and EMC, while the normal ones in odf_main do set those variables | 08:03.28 |
| I guess it would be unlikely to actually cause an interpretation failure, but I'm sure it wouldn't work as well. | 08:03.58 |
chrisl | Ah, okay. Well, if anyone is actually using pdf_cslayer.ps then they can complain, and we can tell them how to handle it properly! | 08:04.31 |
kens | typedef struct single_glyph_list_s {Yes, seems reasonable | 08:04.44 |
| Bah stupid paste again | 08:04.52 |
| I wish I knew what was causing that | 08:05.02 |
chrisl | Some computery thing would be my guess ;-) | 08:05.29 |
kens | You may be right :-) | 08:05.40 |
| I thnk more specifically it may be some keyboardy thing.... | 08:05.55 |
| Possibly due to using a Bluetooth keyboard | 08:06.09 |
Robin_Watts | chrisl: http://www.livenation.co.uk/artist/the-australian-pink-floyd-show-tickets?c=crm_mpbs_AustralianPinkFloyd_030915 ? | 09:27.17 |
pipitas | From my memory the following used to work when having GS do PDF-to-PDF: "-dDownsampleColorImages=true -dColorImageThreshold=1.0 -dColorImageResolution=100". But now it leaves my images untouched. What am I missing? Is my memory failing me and syntax is not correct? | 09:42.45 |
kens | Depends on the resolution of hte images | 09:43.12 |
| ColorImageDownsampleThreshold for starters | 09:44.29 |
pipitas | D'oh! | 09:45.01 |
kens | Thiis in the documentation..... | 09:45.16 |
pipitas | ⦠but this mistake was only here in chat window, as I can see now | 09:46.18 |
kens | I haven't changed anything recently regarding ths, as far as I know it works | 09:46.59 |
| Try an earlier version, you can open a bug report if there's a reproducible problem. | 09:47.36 |
pipitas | Copied from my real command it reads like this: "-dDownsampleMonoImages=true -dDownsampleGrayImages=true -dDownsampleGrayImages=true -dColorImageResolution=100 -dMonoImageResolution=300 -dGrayImageResolution=100 -dColorImageDownsampleThreshold=1.0 -dGrayImageDownsampleThreshold=1.0 -dMonoImageDownsampleThreshold=1.0". | 09:47.57 |
kens | tl;dr | 09:48.09 |
| Talking about it isn't real helpful, if you thnk there's a problem, post a file and a command line, or simply open a bug report with the same informatio. | 09:49.01 |
pipitas | Doesnt work in version from Git for me nor in v9.16. | 09:50.40 |
kens | ^^ | 09:50.40 |
| THs bug http://bugs.ghostscript.com/show_bug.cgi?id=695597 seems to indicate that it worked for me in May. | 09:51.17 |
pipitas | kens: Thx for looking at it â I''m not sure if it isnt a simple f*ckup due to my old age | 09:51.18 |
| The file is from here: https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Linux/7/pdf/SystemTap_Beginners_Guide/Red_Hat_Enterprise_Linux-7-SystemTap_Beginners_Guide-en-US.pdf â complete command line follows: | 09:52.16 |
| gs -dFirstPage=8 -dLastPage=10 -o ../gs-Red_Hat_Enterprise_Linux-7-SystemTap_Beginners_Guide-en-US.pdf -sDEVICE=pdfwrite -dDownsampleMonoImages -dDownsampleGrayImages -dDownsampleGrayImages -dColorImageResolution=100 -dMonoImageResolution=300 -dGrayImageResolution=100 -dColorImageDownsampleThreshold=1.0 -dGrayImageDownsampleThreshold=1.0 -dMonoImageDownsampleThreshold=1.0 ../Red_Hat_Enterprise_Linux-7-SystemTap_Beginners_Gu | 09:52.42 |
kens | That's a 62 page file, SOrry but I'm not going to play with that | 09:52.46 |
| First thing to do; simplify the file (remove extraneous pages) second thng to do; simplify the command line (remove extrabeous parameters) | 09:53.33 |
pipitas | kens: I tried on only 3 of the 62 :-) | 09:53.37 |
kens | You can, of course, debug it yourself and see why downsampling isn't applied (if it isn't) | 09:54.42 |
pipitas | kens: The only "further" way to debug it for now that I can think of is to use a different file which used to work for me. I'll have to search for one first. | 09:56.29 |
kens | You can use a debugger and step through the code, that's what I would have to do and its why I much prefer a simple file and a simple command line | 09:57.04 |
pipitas | kens: but I'm already happy if you do not spot an obvious cmd syntax error for now (I'm getting more and more confused about -sBlah and -dBlah nowadays â thing which I could differentiate blindly and even in the sleep while I was still young⦠:( ) | 09:59.28 |
Robin_Watts | pipitas: A good first stage would be to use mutool clean to extract a single page from the document. | 09:59.53 |
| -sBlah sets a string. -dBlah sets an int. | 10:00.08 |
kens | Not quite, you can set names with -d | 10:00.20 |
| At least, I htnkk you can.... | 10:00.52 |
| One of them certainly takes names | 10:00.59 |
| Yes, -d can accept names as well as numbers | 10:02.24 |
pipitas | "At least, I htnkk you canâ¦." <== makes me so happy that even kens isn't always sure about strange syntax :-) (No malice, gloating or schadenfreude involved at all - just honest relieve!) | 10:10.40 |
kens | I was right though :-) | 10:11.04 |
pipitas | My mnemonic for -sBLah is "s==string", but then: "what is string, what is token??" | 10:11.35 |
kens | strings are strings | 10:12.00 |
| Ths is PostScript we are dealing with, essentially | 10:12.14 |
| Its actually much easier to control the setting of distiller parameters by creating a PostScript file with the ones you want set defined there. | 10:13.30 |
pipitas | But in some cases you can replace "-sBlah=string" by "-dBlah=/string" and it still works⦠That's why I always get thinking what to use now⦠| 10:13.56 |
kens | And tehn you don;t have to remember teh command line syntax. Plus, some parameters can't be set from the command line anyway, other than using -c/-f to send PostScript | 10:13.59 |
| "/<name> isn'a a string, its a name | 10:14.31 |
pipitas | Anyway, I'll take up Robin_Watts' advice to first extract a single page from the 62 and test with that⦠| 10:16.16 |
| Turns out everything works, even with the orig file. Problem is located right in the middle between chair and keyboard. I'm STILL trying to spot the differences between my original cmmand and my current (newly typped) one. I think I'll put them in git and use "git diff" or s.th. â¦. | 10:22.56 |
chrisl | Robin_Watts: I'm not big on tribute groups, tbh | 11:45.00 |
Robin_Watts | chrisl: me either in general, but I've heard really good things about the australian pink floyd. | 12:17.48 |
chrisl | Robin_Watts: d'you know what the ticket prices will be? | 12:20.08 |
Robin_Watts | no idea. | 12:21.26 |
| We have lots of time to think about it :) | 12:21.39 |
chrisl | There's a couple of their shows on youtube - can have a watch, and see what the shows are like | 12:23.42 |
rayjj | kens2: If you look at my email on the web (mail.google.com) it will be shown correctly. There must be something funky with your mail tool. The reply you sent showed the funky characters where the tabs were as well | 15:32.40 |
| kens2: I'll fix up issue 1, but I'd like your input on issue 2 | 15:33.02 |
chrisl | rayjj: your mails are coming through in HTML | 15:33.23 |
rayjj | chrisl: yeah, so? 90% of emails I get are HTML | 15:34.02 |
kens2 | rayjj I don't keep my mail on Google | 15:34.03 |
kens2 | intensely dislikes HTML email, especially because it can display badly | 15:34.28 |
rayjj | kens2: it'll be in your trash on gmail | 15:34.30 |
kens2 | rayjj no it wo't | 15:34.38 |
chrisl | rayjj: 100% of the mails I read are in plain text...... | 15:34.47 |
al__ | hello, I was wondering if muPDF would ever support page labels | 15:35.13 |
| I found a bug report with a proposed patch from over a year ago, but it doesn't look like anything has been done with it | 15:35.50 |
| http://bugs.ghostscript.com/show_bug.cgi?id=695351 | 15:35.52 |
rayjj | kens2: I'll re-send it in text (seems like you need a better mail tool -- I suggest Thunderbird) | 15:35.57 |
kens2 | rayjj I tried thunderbird, it failed to import my mail archive | 15:36.23 |
| Well I cna read you r email now | 15:41.35 |
rayjj | kens2: OK, text version sent (even though it has issue 1, plesae ignore that since you've already relplied) | 15:41.59 |
kens2 | I cna read hte mail now, it'll take me a while. | 15:42.18 |
rayjj | chrisl: BTW, thanks for getting rid of pdf_cslayer.ps | 15:42.36 |
chrisl | rayjj: np - nice that I remembered! | 15:43.05 |
rayjj | kens2: and if you'd rather discuss it here instead of back and forth by email, that's fine | 15:43.14 |
kens2 | WEell the main problem is that in the last 2 months I've forgotten most of what I knew when I wrote the code | 15:43.40 |
rayjj | I did email because it was too much for IRC and I didn't want you to have to read the logs | 15:44.02 |
kens2 | I don't think you can change the allocation, though probably I don't understand your issue properly. The new child must be the same size as the source device | 15:44.14 |
rayjj | kens2: sort of like post traumatic amnesia ? ;-) | 15:44.32 |
kens2 | Well like Robin I have a poor memory | 15:44.44 |
| And in the last couple of moths I have been over all kinds of areas of the code, right now I'm meddling with PJL and I really don;'t remember the subclassig stuff that well now | 15:45.16 |
| I'd be *extremely* wary of changing any of the allocatoin stuff surrounding the device subclassing, it was really hard to get ths working and I'd want to thnk very carefullyt again before changing it. | 15:46.20 |
| Its true that the newly created subclassing devices currently don't maintain GC pointers, I was sort of hoping that in general they wouldn't | 15:46.53 |
| I still don;t thnk I understand your point 2, can you reference me a point in the code to look at ? | 15:48.42 |
chrisl | rayjj: If I'm understanding what you're saying, you will need to re-factor spotcmyk so that it is smaller that the device it might be replacing | 15:49.44 |
kens2 | Ah, going by what chrisl says, then yes. subclassing devices **must** be minimal | 15:50.34 |
rayjj | kens2: so it looks like the existing subclassing devices are all really simple -- don't do anything except finagle procs | 15:50.54 |
chrisl | may be misunderstanding..... | 15:51.03 |
kens2 | They are required to fit into the smallest possible device, which is why they carry round their own memory area. | 15:51.11 |
| rayjj that is essentially true, and the main reason for using such a device | 15:51.48 |
Robin_Watts | kens2: I'm convinced that *everyone* has a bad memory. Just some people don't realise it... | 15:51.58 |
kens2 | Robin_Watts : then I'm ahead of the game :-) | 15:52.15 |
chrisl | Everyone has a bad memory *for at least something*...... | 15:52.58 |
kens2 | rayjj so your problem seems to be, from my reading and chris's hint, that you want to use an existing rendering device as a subclassing device, and that device is larger than the existing rendering device ? | 15:54.52 |
rayjj | basically, it sounds like the device subclassing can't really do what would be needed to 'render off to the side' then pass the data along after, as that requires memory (unless everything is done in non_gc_memory, which may be OK) | 15:55.39 |
kens2 | rayjj it can, but only if the subclassing device is no large than the smallest possible device | 15:56.05 |
| I thought I mentioed ths in the documentation somewhere | 15:56.24 |
rayjj | kens2: or if all of the extra stuff needed is in the "private' data | 15:56.41 |
kens2 | rayjj exactly, that's the main point of the private pointer | 15:57.03 |
| rayjj ths is in the docs | 15:58.21 |
| see subclass.htm, "new device members" paragraph 2 | 15:58.36 |
rayjj | kens2: hmm... I hadn't noticed there was a doc. It probably should be added to the 'developers' section of the Readme.htm | 16:10.25 |
kens2 | rayjj henrys asked me to cast the PDF into HTML, so I did. Feel free to put a pointer to it in the developers section of the readme | 16:10.59 |
rayjj | kens2: I was trying to puzzle things out by the .c and .h files (which is usually better anyway given how we are at maintaining docs) | 16:11.41 |
kens2 | Well I did send the documentation round as a PDF file when I first asked people to review the code | 16:11.58 |
| THe HTML is the same thing reformatted | 16:12.07 |
rayjj | kens2: but I will read the doc to see if it looks like using subclassing is easier than doing a compositor (or perverting the pdf14 device) | 16:12.27 |
kens2 | rayjj I have no idea, but be aware that you should test whatever you do with, at least, the FirstPage and LastPage devices in place. | 16:13.02 |
| There's a way to force the internal classes ot be present and active, even without the command line parameters, which should make it easy, becasue that way clister runs will do the testing for you | 16:13.46 |
| In gdevkrnlsclass.c change the #define FORCE_TESTING_SUBCLASSING to 1 | 16:14.21 |
| That will cause 2 devices to be pushed into the chain, and when running a few of the PCL tests it will also push the momochrome palette device for a total of 3 devices | 16:15.12 |
rayjj | kens2: yes, I saw that in my code perusal, and have stepped through the process with the debugger | 16:25.07 |
| I'll go ahead and fix the VMerror issue #1 and the Readme.htm if that's OK with you. Do you want to review the changes before pushing to origin? | 16:26.18 |
kens2 | rayjj no the changes are simple, its an obvious oversdight | 16:35.30 |
| I'm off now, family is home, goodnight all | 16:35.50 |
amish | does mupdf support rendering a pdf file while downloading? | 20:50.52 |
sebras | amish: the mupdf sample application (named mupdf-x11-curl) for linux supports this. | 21:32.56 |
| amish: the mupdf library itself doesn't really fetch pdfs over e.g. http, but the sample applications illustrates that it can be done in an application if you implement it. | 21:33.59 |
| amish: this is useful if you e.g. use the mupdf library to implement an application of your own. | 21:36.12 |
mvrhel_laptop | cool. I have some basic interactive annotations things working now in gsview (just push buttons for now) | 23:19.19 |
| more work to do, but I have the basic plumbing in place | 23:19.57 |
| done for the day for a bit | 23:21.00 |
| Forward 1 day (to 2015/09/04)>>> | |