| | <<<Back 1 day (to 2012/11/08) | 2012/11/09 |
| Robin_Watts_ | Ch3rrryC0ke: page->mediabox contains the intersection of cropbox and mediabox, which is what you need, I think. | 00:23.38 |
| | Hmm. | 00:25.48 |
| | fz_bound_page is no good. | 00:26.04 |
| | You can use ((pdf_page *)page)->mediabox for now, if you #include "mupdf-internal.h" | 00:27.08 |
| | I think that's a missing thing in our API. tor8, sebras: Are you following this? | 00:27.31 |
| Ch3rrryC0ke | I need the difference between cropbox and mediabox | 00:35.38 |
| | e.g. the "left margin" by getting cropbox.left - mediabox.left | 00:35.53 |
| Robin_Watts_ | Right, so use ((pdf_page *)page)->mediabox to get cropbox. | 00:43.44 |
| | Then use fz_bound_page to get mediabox. | 00:44.11 |
| | then you can subtract the two to get the margin. | 00:44.38 |
| | actually, no. Just use ((pdf_page *)page)->mediabox.{x0,y0} | 00:45.22 |
| | as fz_bound_page always returns a box rooted at 0,0. | 00:45.36 |
| | I think we need to add an API to get page margins that will return that. | 00:45.54 |
| | I need to discuss that with tor8 and co. | 00:46.04 |
| Ch3rrryC0ke | ah gotcha thanks! | 00:46.29 |
| sebras | http://www.bttr-software.de/forum/forum_entry.php?id=9530 | 08:23.44 |
| kens | OMG..... | 08:24.36 |
| sebras | kens: :) | 08:24.41 |
| kens | I thought people complaining about GS on OS/2 was bad.... | 08:25.03 |
| sebras | I found this some time ago but forgot it in the bookmark list on my phone. | 08:25.43 |
| | I wonder what they did with the makefiles. | 08:26.01 |
| chrisl | Since that's all clearly the restul of some group delusion, I'm not sure I want to know what they did with the makefiles....... | 08:26.52 |
| sebras | I downloaded the source just to take a look, and they have definitely been faithful to the cause: | 08:29.11 |
| | drwx------ 2 sebras sebras 4096 Feb 18 2011 FITZ | 08:29.14 |
| | drwx------ 3 sebras sebras 4096 Feb 18 2011 FONTS | 08:29.16 |
| | drwx------ 7 sebras sebras 4096 Feb 19 2011 LIBS | 08:29.19 |
| | -rw-r--r-- 1 sebras sebras 9562 Feb 19 2011 MAKEFILE | 08:29.21 |
| | everything is in uppercase as expected. :) | 08:29.46 |
| chrisl | I was jokingly discussing whether a GS port to CP/M was on the cards a while back...... | 08:32.36 |
| kens | Old Operating Systems never die.... | 08:32.59 |
| chrisl | No, they just keep running Ghostscript...... | 08:33.29 |
| sebras | hopefully gs never has to be bug-backwardscompatible, like the sw at work. :-( | 08:35.35 |
| chrisl | I've seen at least two complaints about that since I joined! We don't have to be, but some folks want that :-( | 08:36.43 |
| sebras | looks like the djgpp-guys fix some compiler warnings too back in 0.7... I wonder if those this apply to 1.1... | 08:38.33 |
| | anyway time to go. | 08:38.56 |
| Robin_Watts_ | tor8: Morning | 12:51.30 |
| tor8 | Robin_Watts_: hey | 13:06.15 |
| Robin_Watts_ | We found a couple of bugs in the tagged release. | 13:08.42 |
| paulgardiner | Robin_Watts_: Turned out there was two bugs, but both fixed now, I think. | 13:08.54 |
| Robin_Watts_ | paulgardiner has fixed all but one, and he's looking at that now. | 13:08.57 |
| | oh, right, cool. | 13:09.00 |
| | I'll pull and rebuild. | 13:09.07 |
| tor8 | Robin_Watts_: want to re-tag? | 13:10.11 |
| paulgardiner | The problem that caused the overwritten text, although I was able to fix it in the android code, may be a sign that there's a bug in fz_new_draw_device_with_bbox | 13:10.18 |
| Robin_Watts_ | tor8: I'll retag locally and push the tag when I've tested. | 13:10.29 |
| tor8 | we don't use signed tags so changing it is trivial (but may cause confusion) | 13:10.30 |
| Robin_Watts_ | paulgardiner: Is this a good point to add recycling of bitmaps? | 13:10.58 |
| paulgardiner | Robin_Watts_: I was thinking not, just in case. | 13:11.12 |
| Robin_Watts_ | As far as I can see, whenever we do a bitmap = null; we should really do bitmap.recycle(); bitmap = null; | 13:11.19 |
| | Fair enough. | 13:11.27 |
| paulgardiner | That may not quite be true. Something else may have a reference to the bitmap | 13:11.54 |
| Robin_Watts_ | oh, right, yes. | 13:12.05 |
| paulgardiner | fz_new_draw_device_with_bbox(ctx, pix, box) seems to require that box is contained in defined_rect_of(pix), which is easy enough to arrange, but I'd have expected not to be ncessary. | 13:13.58 |
| Robin_Watts_ | paulgardiner: Ah, so fz_new_draw_device_with_bbox should insersect box with the bbox of pix really. | 13:15.19 |
| | And that would render your second patch unnecessary ? | 13:15.37 |
| paulgardiner | I'd have thought so. | 13:15.49 |
| Robin_Watts_ | Let's go with your fix for now. | 13:17.38 |
| paulgardiner | It's a little unclear, because if there is no intersection at all then you are allowing the caller to create a device that wont draw anything | 13:17.46 |
| | Robin_Watts_: my fix is needed anyway for efficiency | 13:18.00 |
| Robin_Watts_ | yeah, but that should be permissible I guess. | 13:18.09 |
| paulgardiner | it was a bug in my code too, just it probably shouldn't have caused rendering artifacts | 13:18.30 |
| Robin_Watts_ | tor8: Did you see the conversation I had here last night with Ch3rrryC0k3? | 13:18.54 |
| | I think there is a hole in our API; he gets a page and uses fz_bound_page to get a rectangle for it, and then positions annotations on it. | 13:22.53 |
| | but when he saves them back and rerenders they can be shifted. | 13:23.04 |
| | The problem is that the page has a MediaBox of [0 0 W H] and a CropBox of [X Y W H] | 13:23.41 |
| | so pdf_bound_page returns [0 0 W-X H-Y] and he positions annotations accordingly. | 13:24.08 |
| | Really he needs some way to get the margin values out. | 13:24.17 |
| | And he's a paying customer :) | 13:24.39 |
| | I gave him a workaround. | 13:24.55 |
| tor8 | Robin_Watts_: read bits of it | 13:30.32 |
| | Robin_Watts_: right; did you make him look at the /MediaBox and /CropBox dict entries directly instead? | 13:30.55 |
| | it gets even hairier if the page has a /Rotate entry | 13:31.18 |
| Robin_Watts_ | How does he (legally) get from a page* to those if he's opened it using fz_ ? | 13:31.40 |
| | I said to do ((pdf_page *)page)->mediabox and look at that. | 13:31.57 |
| | but that means he has to use #include "mupdf-internal.h" | 13:32.10 |
| tor8 | Robin_Watts_: well, he's adding annotations by creating pdf_objs ... I don't think he should be using the fz_page abstraction level then | 13:32.22 |
| Robin_Watts_ | I think he uses a different tool to add the annotations. | 13:32.36 |
| | iText? | 13:32.40 |
| tor8 | he ought to look at the xref->pageobj[] entries | 13:32.43 |
| Robin_Watts_ | Is there a reason why we always return a page bbox rooted at (0,0) ? | 13:33.24 |
| tor8 | the fz_page does "normalization" of oddities like mediaboxes, rotated coordinate systems, dpi differences with /Units in PDF, and XPS being based on 96dpi, etc | 13:33.33 |
| | so that you always know that the fz_page has the origin at 0,0 and a coordinate system at 72 dpi | 13:34.02 |
| | before that, we had the viewers do all sorts of funky format-specific math | 13:34.17 |
| Robin_Watts_ | tor8: I wonder if we really need an fz_page_margins call then. | 13:34.18 |
| tor8 | Robin_Watts_: why? xps has no such thing, and there are two dozen different boxes in PDF | 13:34.43 |
| | mediabox, cropbox, artbox, bleedbox, whateverbox | 13:34.54 |
| Robin_Watts_ | or a call to get a transformation matrix to go from fz space to document space. | 13:35.15 |
| | That sounds nicer, I think. | 13:35.44 |
| tor8 | Robin_Watts_: we could add a pdf_page->transform that has the device independent space -> document space transform | 13:36.42 |
| | which would also let him deal with rotations | 13:36.50 |
| Robin_Watts_ | yeah. | 13:36.55 |
| tor8 | oh wait, we already do. pdf_page->ctm | 13:37.16 |
| | that takes care of both rotations and UserUnit | 13:37.58 |
| | so an inverted pdf_page->ctm will get him from fitz space back into pdf document space | 13:38.56 |
| Robin_Watts_ | tor8: How can he get that from the fz_ level ? | 13:39.42 |
| | paulgardiner: Something I've noticed with your highlighting stuff. | 13:41.35 |
| | I bring up a document, and click on a field. | 13:41.45 |
| | The text entry box pops up, and I fill it in, click OK, and it goes back into the document with the field highlighted. | 13:42.14 |
| | Then I click on another field, and the text entry box pops up. Confusingly the document in the background leaves the old field highlighted until the text entry box is closed. | 13:42.48 |
| | I guess I understand why, but it's confusing. | 13:43.09 |
| Robin_Watts_ | is playing the Majid role, here, sorry. | 13:43.20 |
| paulgardiner | Hmmm, I didn't think it did that. | 13:46.01 |
| | The change over isn't instantaneous, but I thought it did change to the focussed field | 13:47.26 |
| | I thought I'd sorted that in the Windows build too. | 13:47.46 |
| Robin_Watts_ | I'm using android, with the fda-1571.pdf file | 13:48.18 |
| | And now it's working. | 13:48.46 |
| | So maybe ignore me. | 13:48.52 |
| | yeah, sorry. | 13:49.12 |
| paulgardiner | Hopefully it was the update delay that was confusing. | 13:49.19 |
| Robin_Watts_ | I meant to say, the buttons, and radiobuttons and checkboxes look really nice on the files I've tried. Nice 3d effect. | 13:50.00 |
| | Did you do that ? | 13:50.04 |
| paulgardiner | I have to admit no. Those are already in the files. Well, other than AcroForm_BasicToggle.pdf | 13:52.23 |
| Robin_Watts_ | Can we steal them? :) | 13:52.48 |
| paulgardiner | :-) | 13:53.16 |
| tor8 | Robin_Watts_: you can't, because you don't know that you have a pdf | 13:53.28 |
| | Robin_Watts_: but if you *do* know you have a pdf_document, you can cast (pdf_page*)my_fz_page | 13:53.48 |
| Robin_Watts_ | OK. Should we have a fz_page_transform(); API entry to retrieve it ? | 13:54.10 |
| tor8 | Robin_Watts_: NO! | 13:54.16 |
| | it's bloody format specific. maybe add a pdf_page_device_transform. | 13:54.49 |
| Robin_Watts_ | Or maybe at least we should have a way of identifying the document type so we know that it's safe to cast. | 13:55.05 |
| paulgardiner | Robin_Watts_: Please carry on being Majid. These bug reports are helping hugely. | 13:56.16 |
| tor8 | Robin_Watts_: an ugly workaround for that is to test if (doc->close == pdf_close) | 13:56.27 |
| Robin_Watts_ | Maybe a meta call to get a pdf_page * from an fz_page * or NULL if it's not a pdf, or something. | 13:56.45 |
| tor8 | Robin_Watts_: but really, if he's doing pdf specific stuff, he can use the pdf_document level directly | 13:56.53 |
| | but that does mean using the -internal headers | 13:57.00 |
| Robin_Watts_ | tor8: Right, but it would be nice to have a way to jump from the fz level to the pdf level legally. | 13:57.16 |
| tor8 | Robin_Watts_: a safe downcast for each document type perhaps? | 13:57.29 |
| Robin_Watts_ | tor8: right. | 13:57.39 |
| tor8 | pdf_document *pdf_cast_document(fz_document *doc) | 13:57.48 |
| | that can test the function pointers and return a cast pointer or NULL | 13:58.07 |
| Robin_Watts_ | yes, that sort of thing. | 13:58.14 |
| | The way you've just done it there doesn't bloat the fz level. | 13:58.32 |
| tor8 | I could live with that, but outside of this one customers specific needs I don't really see much use for it | 13:58.35 |
| Robin_Watts_ | and it gets to be hidden in the mupdf-internal.h layer. | 13:58.49 |
| | lunchtime. | 13:59.45 |
| | http://ghostscript.com/~robin/MuPDF-1.1-forms-tech-preview | 16:21.23 |
| | tor8, paulgardiner,sebras: ^ | 16:21.37 |
| | Those are the built binaries. | 16:21.44 |
| | tor8: I haven't updated the website, because I believe it's in your public_html on casper, and I don't have write access without resorting to sudo. | 16:22.51 |
| | kens, chrisl: Can either of you think of a way to get CMYKcm files out of gs ? | 16:24.27 |
| kens | Basically, no | 16:25.06 |
| | we don't produce light cyan and magenta | 16:25.14 |
| Robin_Watts_ | I guess I could take a CMYK file and extract the light parts of CM to cm planes, and rescale. | 16:25.24 |
| kens | These are usually generated form the CYan and Magenta by the CMS | 16:25.37 |
| | You might be able to use the CMS to do this, but I'm not certain and you'd need a special profile | 16:26.14 |
| Robin_Watts_ | kens: Yeah, I was wondering if there was some magic that could be wrought with color profiles. | 16:26.17 |
| kens | You'd have to ask Michael about that, its beyond me | 16:26.43 |
| Robin_Watts_ | Me too. | 16:27.05 |
| henrys | wow chrome segv's reading gmail. | 16:34.39 |
| Robin_Watts_ | Hi henrys. Feeling better? | 16:35.38 |
| henrys | yes much better thanks | 16:38.46 |
| LeastOne | Hello all. I'm wondering if someone might be able to help with the following problem⦠| 16:40.29 |
| henrys | amazing how effective the immmunity system is at evicting stuff when something goes is awry. | 16:40.33 |
| LeastOne | I'm using Batik (for SVG->PDF) to create PDFs. I have a problem where Batik is converting grayscale images into RGB images during the conversion to PDF. I then have a follow-up process that runs the PDF through Ghostscript (to convert to CMYK). | 16:40.38 |
| | There does not appear to be a way to instruct Batik to leave the grayscale images alone. Is there any way to instruct Ghostscript to detect that an RGB image is actually grayscale compatible (because all three channels R, G and B will be equal) and convert the image back to grayscale or CMYK with only K specified? | 16:40.54 |
| Robin_Watts_ | Are we expecting ray and/or michael today? | 16:53.30 |
| | ah, ray was here earlier. | 16:53.42 |
| henrys | I usually hear form michael when he isn't coming in. It's not that late yet. | 16:54.46 |
| Robin_Watts_ | I know there is an ICC meeting on "at some point soon" | 16:55.27 |
| kens | LeastOne no there is no provision for that | 16:57.03 |
| LeastOne | Yea, I didn't suspect there would be ... I'm open to suggestions if anyone has another idea that comes to mind? | 16:58.20 |
| chrisl | marcosw: ping | 17:03.14 |
| marcosw | chrisl: morning | 17:03.22 |
| chrisl | marcosw: can you expel users from bugzilla? | 17:03.45 |
| marcosw | I think so, but there isn't anything to stop them for signing up with a new account. | 17:04.21 |
| chrisl | marcosw: we seem to have a little spam appeared (see attachment): http://bugs.ghostscript.com/show_bug.cgi?id=690000 | 17:04.42 |
| marcosw | chrisl: annoying; how many bugs does this affect? | 17:05.54 |
| chrisl | So far only one, but I thought it would be good to avoid any more being affected | 17:06.17 |
| marcosw | okay, I'll ban the user in question and delete the attachment. | 17:07.14 |
| henrys | spamming bugzilla ... they've open a new attack front. | 17:16.24 |
| kens | Heading off now, ahev a good weekend everyone. | 17:17.32 |
| chrisl | henrys: I'm putting off pulling the gs_fapi branch onto master until Monday - I've made some fairly large Makefile changes/additions, and I don't want to commit and disappear for the weekend, just in case. | 17:18.17 |
| henrys | okay no big deal | 17:19.00 |
| chrisl | I've added ???? to our build system, so it will work like the other libs (luratech etc) | 17:19.35 |
| henrys | that will be much better, take care to handle proprietary stuff right, it is easy to make a mistake. | 17:22.37 |
| | when moving things around, that is. | 17:23.35 |
| chrisl | Other than the list of source files (which we can't really avoid), there's nothing that wasn't already in our makefiles | 17:24.37 |
| henrys | ugh that really is unavoidable? | 17:26.19 |
| chrisl | I'd be interested to see a makefile that doesn't rely on it! ;-) | 17:27.20 |
| henrys | well you would use an include and have different makefiles included based on a ???? build vs. normal. | 17:28.24 |
| | the ???? makefile would live in a private repo | 17:28.51 |
| chrisl | Yeh, I could do that. We don't do that for Bitstream or Luratech, though | 17:29.12 |
| henrys | Let's see how it goes with what you have. | 17:29.35 |
| chrisl | I can always ask Karen if they'd have a problem with it | 17:30.07 |
| henrys | and unavoidable | 17:33.09 |
| chrisl | I'll see if I can put the ???? makefile in the ???? directory in svn-private. The question is whether a make "include" directive can use a variable for the path | 17:34.43 |
| henrys | Yeah I see you guys on google alert all the time. | 17:37.15 |
| Robin_Watts_ | ok, give it 5 mins to regenerate the logs. | 17:39.40 |
| chrisl | Ah, cool, henrys I can have the makefile for "that library" only ship with the source for "that library", and pull in the make file from there. | 17:40.14 |
| henrys | nice I like that better. | 17:41.58 |
| chrisl | I'll do it Monday, it won't take long | 17:42.11 |
| Robin_Watts_ | tor8: Aha. | 17:43.01 |
| | I'm footling with the mupdf website. | 17:43.17 |
| | http://ghostscript.com/~robin/MuPDF-1.1-forms-tech-preview | 17:43.24 |
| | Those are the binaries for the release. | 17:43.31 |
| | Give me 20 minutes and you can "ew!" at my html 5k1llz. | 17:43.56 |
| tor8 | Robin_Watts_: how do you want to do the release? add an item to the "news" page with a pointer to that directory? | 17:44.44 |
| | or upload to google code? | 17:44.54 |
| Robin_Watts_ | I am writing a release page now. | 17:45.09 |
| | Where we point the downloads is up to you, but those are the binaries. | 17:45.24 |
| henrys | I never knew git took uncommitted code from master to a branch without stash push and pop. An accidental discovery | 17:47.00 |
| Robin_Watts_ | henrys: With git you have a working set, the index, and the set of commits. When you checkout to a new branch, it changes all the files that changed between the commit you are on, and the commit you are going to. | 17:48.47 |
| | If you have changes in your working set in files that intersect with the files that commit would checkout, it will complain at you. | 17:49.12 |
| | If not, it will leave your changes in the working set alone. | 17:49.26 |
| | If it trashed your changes in the working set, you could easily lose work, and git tries very very hard to avoid that. | 17:49.54 |
| | Indeed, I realised something recently... One of the few things you ever have to do that has potential to lose work is to: git reset --hard | 17:50.29 |
| | So instead don't do that. Use git stash && git stash drop | 17:50.45 |
| | That does exactly the same thing, but leaves a trace in the reflog so you can get it back for 30 days in case it was a mistake. | 17:51.12 |
| | Aha. An mvrhel_laptop. | 17:51.39 |
| | Morning mvrhel_laptop. Are you going to the ICC meeting that Max was talking about ? | 17:51.57 |
| mvrhel_laptop | Hi Robin_Watts_ no I was not planning to attend. Since it is in LA, I was wondering if Ray should go | 17:53.00 |
| | Robin_Watts_: thanks for answering all those questions | 17:53.16 |
| chrisl | Ray mentioned going to the ICC thing, since it's local | 17:53.25 |
| Robin_Watts_ | mvrhel_laptop: Please, if you see me say anything stupid, speak up! | 17:53.45 |
| mvrhel_laptop | I used to have a nice way to generate CMYKcm for a real printer that I had characterized years ago | 17:53.50 |
| | It uses 1-D LUTs to generate C'M'cm from CM | 17:54.13 |
| | which is the way they do it usually | 17:54.50 |
| ManFree | A newbie question... Assuming I have an input PDF or PostScript document, I want to extract part of it (a box by coordinates), rotate that box, and place it on a new document. Is this possible? I came here because it seems External XObjects are the way to go, but I am not sure they are supported in Ghostscript... | 17:54.56 |
| mvrhel_laptop | Robin_Watts_: If you want me to generate some raw CMYKcm 8 bit contone images using the tool I can do that. The images would be in a raw format though. Or I can just give you the tables and we can add it as another parameter. That may make the most sense | 17:56.33 |
| Robin_Watts_ | So we'd read CMYK in, expand it to CMYKcm internally, and output... what ? | 17:57.09 |
| mvrhel_laptop | Robin_Watts_: we would need to figure out how best to display. I would suggest that we start having and option to output NChannel planar raw | 17:57.28 |
| | so that we can view it in Photoshop. Actually PSD would be nice | 17:57.48 |
| Robin_Watts_ | I am tempted to say we should import CMYKcm and output CMYKcm (as pam's) | 17:57.51 |
| mvrhel_laptop | with a mapping for the dot color | 17:57.53 |
| Robin_Watts_ | mvrhel_laptop: So *you* can view it you mean :) | 17:57.59 |
| mvrhel_laptop | well yes. and those with photoshop | 17:58.10 |
| Robin_Watts_ | I only have photoshop elephants. | 17:58.21 |
| | That only does RGB. | 17:58.33 |
| henrys | ManFree:here is a start http://stackoverflow.com/questions/6183479/cropping-a-pdf-using-ghostscript-9-01 you might want to continue the thread there. | 17:58.54 |
| mvrhel_laptop | Robin_Watts_: OK. If you are fine with it, I may add the capability to the project to do psd n channel for the output | 17:58.56 |
| Robin_Watts_ | mvrhel_laptop: I certainly have no problem with that! | 17:59.06 |
| ManFree | henrys: Thakns, I am going to read that | 17:59.12 |
| mvrhel_laptop | ok. question then is what about the input.... | 17:59.25 |
| Robin_Watts_ | I'd love to find a way that doesn't tie people into to owning Adobe's crap though. | 17:59.28 |
| henrys | ManFree:not to push you off, but your work might be of use to others and it will be archived there. | 17:59.37 |
| Robin_Watts_ | If we're going to output to .psd, then inputting from .psd wouldn't be awful. | 17:59.58 |
| mvrhel_laptop | Robin_Watts_: true | 18:00.08 |
| | generally I like to work in raw planar a lot though | 18:00.27 |
| | for input and output. the down side of that for the output is that I can't control the dot color | 18:00.47 |
| Robin_Watts_ | mvrhel_laptop: If only there was a free image viewer that could read that though. | 18:00.55 |
| mvrhel_laptop | like I can with psd | 18:00.57 |
| | Robin_Watts_: that would be easy enough for us to make one.... | 18:01.09 |
| | isn't there a multichannel ppm format? | 18:01.26 |
| | I guess it doesnt specify dot color through | 18:01.41 |
| Robin_Watts_ | Yes. pam with new tupltype. | 18:01.45 |
| mvrhel_laptop | how is that displayed? | 18:01.52 |
| Robin_Watts_ | At the moment I spit out PAMs with a "TUPLTYPE CMYK" line. | 18:02.16 |
| | and imagemagick converts that to pngs. | 18:02.32 |
| | We could easily spit out "TUPLTYPE CMYKcm" and have our own viewer read that. | 18:02.53 |
| mvrhel_laptop | oh. what does imagemagik do with more than 4 or noncmyk tupletype | 18:02.55 |
| | Robin_Watts_: oh ok | 18:03.04 |
| Robin_Watts_ | It probably barfs :( | 18:03.11 |
| mvrhel_laptop | is there a way to pack in any extra information in the pam type? | 18:03.40 |
| Robin_Watts_ | but it may be possible to tell it to extract individual planes. | 18:03.45 |
| mvrhel_laptop | like comments | 18:03.46 |
| Robin_Watts_ | Comments, yes. | 18:03.50 |
| mvrhel_laptop | ok. so we could leverage that for dot color | 18:04.00 |
| Robin_Watts_ | # WHATEVER YOU WANT | 18:04.01 |
| mvrhel_laptop | information | 18:04.02 |
| Robin_Watts_ | Yeah. | 18:04.04 |
| mvrhel_laptop | so there is our non adobe solution | 18:04.36 |
| Robin_Watts_ | cool. | 18:04.43 |
| ManFree | henrys: Just went over it (yes, I will publish the script once I figure it out :) ), the next step would be to embedd several such cropped PDFs into a single document, each embedding is with rotation. What PS/PDF/GS commands should I be reading about? | 18:04.45 |
| mvrhel_laptop | we just need to write a viewer for it.... | 18:04.55 |
| | you should have that finished tonight | 18:05.12 |
| | we would want the capability in the viewer to turn off and on individual planes | 18:06.04 |
| Robin_Watts_ | I could knock one up for RISC OS, no problem. | 18:06.12 |
| | I've never written a Win32 desktop app. | 18:07.27 |
| mvrhel_laptop | I can port it over to that | 18:07.33 |
| | I have done a few image viewer application there | 18:08.11 |
| | ones that included CMYKcm even... | 18:08.26 |
| Robin_Watts_ | (The RISC OS line was a gag) | 18:08.30 |
| mvrhel_laptop | hehe | 18:08.35 |
| | ok. well for now, let me add in the photoshop option and lets think more about the pam case | 18:09.14 |
| Robin_Watts_ | ok. | 18:09.30 |
| mvrhel_laptop | I am sure the customer has photoshop | 18:09.32 |
| Robin_Watts_ | I could get Miles to buy me it. That might be the simplest way. | 18:09.49 |
| mvrhel_laptop | Yes. That would probably be best | 18:10.10 |
| Robin_Watts_ | $700 ! | 18:10.35 |
| mvrhel_laptop | The hours we would spend building our cool viewer application and pushing our new standard might pay for it | 18:10.44 |
| | wow | 18:10.52 |
| | that is crazy. maybe we do need to build a viewer application and start selling it | 18:11.12 |
| | or add the capability to GIMP | 18:11.27 |
| | I wonder why no one has done that | 18:11.48 |
| Robin_Watts_ | GIMP is hardwired to RGB, I think. | 18:11.51 |
| mvrhel_laptop | ouch | 18:12.02 |
| Robin_Watts_ | I think it can load stuff and convert to RGB as it loads,but bang goes turning layers on or off. | 18:12.12 |
| mvrhel_laptop | ok | 18:12.27 |
| Robin_Watts_ | I wonder if I could code it in javascript using the canvas... | 18:12.29 |
| Robin_Watts_ | slaps self. | 18:12.37 |
| mvrhel_laptop | The other option is the following. We write a little tool that will take the ntuple pam and converts to RGB | 18:13.37 |
| | then anyone can view that | 18:13.50 |
| Robin_Watts_ | Ah, with the excellent US -> UKP pricing, it's only 630UKP. Geez. | 18:14.02 |
| mvrhel_laptop | or we add that capability in the halftone tool | 18:14.03 |
| | actually the later may not be a bad idea. Add in the a little option to read in the dot colors to use | 18:14.35 |
| Robin_Watts_ | We could, but I fear with CMYKcm we REALLY want to be able to toggle layers on or off. | 18:14.36 |
| mvrhel_laptop | Robin_Watts_: yes | 18:14.45 |
| | that is handy | 18:14.47 |
| Robin_Watts_ | I really do wonder about a web viewer... | 18:15.40 |
| mvrhel_laptop | Robin_Watts_: That I would not know how to code up. So hat be in Javascript | 18:16.11 |
| | s/hat/that would | 18:16.22 |
| Robin_Watts_ | Doing stuff in a browser has the advantage that the UI just works everywhere. | 18:16.26 |
| | yeah. | 18:16.27 |
| mvrhel_laptop | right. I can see where that might be nice | 18:16.46 |
| Robin_Watts_ | I'll ponder that over the weekend. | 18:17.00 |
| mvrhel_laptop | ok. you might be on your own with that one. I am not a javascript guy. | 18:17.22 |
| | but I would try to help where you needed it | 18:17.48 |
| | in the meantime I will had the psd support | 18:17.58 |
| | a/had/add/ | 18:18.02 |
| | can't type today. | 18:18.07 |
| | I will call ray to see if he is going to the icc meeting | 18:18.39 |
| Robin_Watts_ | actually... if we did a javascript viewer, it would be pretty trivial to allow us to show the different levels of the dots separately. | 18:19.12 |
| mvrhel_laptop | cool | 18:19.22 |
| henrys | ManFree:I was hoping you would post the problem on stack overflow with a complete description of your problem and it will get answered there. But if you want reading references and are going to use the "-c postscript and pdfmark" stuff like Kurt you should read about transformations, setpagedevice, and pdfmark in the postscript and pdf manual (google for those manuals), good luck. | 18:20.07 |
| Robin_Watts_ | tor8, henrys: So where are we going to release the mupdf tech preview from? | 18:21.42 |
| | downloads.ghostscript.com/public/1.1-forms-tech-preview ? | 18:22.09 |
| henrys | I was hoping you would hang a link to it on the release page? | 18:22.12 |
| Robin_Watts_ | henrys: Yes, I'm doing that now. See http://mupdf.com/index-forms.html | 18:22.36 |
| | The question is where I make the "here"'s point to. | 18:23.22 |
| tor8 | Robin_Watts_: I changed perms so you should be able to edit mupdf.com without sudo | 18:24.38 |
| henrys | I was hoping we could something at ghostscript.com/download parallel to mupdf | 18:24.43 |
| | MuPDF Forms 1.1a | 18:25.20 |
| tor8 | Robin_Watts_: (also, no need for .html suffix on the urls. apache is sometimes smart.) | 18:25.45 |
| henrys | then whe you click on that it would give you the banner and downloads - there is no need to distinguish between commercial and public. | 18:26.25 |
| | yet | 18:26.36 |
| Robin_Watts_ | I think we'd want to have any downloads at least pass through http://mupdf.com/forms.html | 18:26.37 |
| henrys | so the MuPDF Forms 1.1a link on ghostscript.com/download points to that. That's what I meant by "banner" I guess it is more than a banner. | 18:28.50 |
| mvrhel_laptop | bbiaw | 18:28.59 |
| Robin_Watts_ | Right. Certainly having a banner point from there is fine. | 18:29.26 |
| | I mean a link on ghostscript.com/download | 18:29.45 |
| | But we actually need a place to put the binaries. They could go on google code, or they could go on downloads.ghostscript.com in a subdirectory in public. | 18:30.15 |
| | I have no real preference. | 18:30.22 |
| | But I need to know where, so I can fix the links. | 18:30.33 |
| henrys | thats what I'm saying add a link on ghostscript.com/download MuPDF Forms 1.1a directly underneath MuPDF 1.1 that points to http://mupdf.com/forms.html | 18:30.58 |
| Robin_Watts_ | Yes, I agree. I will do so (or will hassle chrisl into doing so if I can't see how) | 18:31.22 |
| henrys | hi norbertj just regression testing your dl font patch. | 18:54.38 |
| norbertj | Hello Henrys, was hoping for that ;) So if all is ok, we can get a patch out to the customer (was ringing us every day ) | 18:55.56 |
| henrys | very nice catch in the font selection code | 18:56.36 |
| | ^^^ norbertj | 18:57.13 |
| | norbertj:we should probably give you access to our regression test if you are going to start fixing the hard ones. Let me know if you are interested. | 18:58.11 |
| norbertj | yes, was lucky with that. I allready found that the selected_id was being reset, so simplifying the testfile and getting the breakpoints at the correct locations eventually did. | 18:58.17 |
| henrys | I thought for sure the macros were interacting is some odd way and suspected a long slog to find the problem. | 18:59.44 |
| | norbertj:we do have some regressions though, the pcl dragon doesn't usually yield that easily ;-) | 19:00.47 |
| norbertj | Most of the macros (i.e. id 0) was just used to add in a comment (printing BARcode such and so) etc. And the other macro 99xx we don't have (probably installed on the printer of the customer). | 19:01.08 |
| Robin_Watts_ | henrys: How about that? http://ghostscript.com/download/ | 19:01.11 |
| norbertj | For easier debugging I suspect, but it confused me at first also. | 19:01.45 |
| Robin_Watts_ | tor8: Are you happy for me to rename index-forms.html to index.html ? | 19:02.46 |
| norbertj | henrys, b.t.w. the customer has a number of our faster printers installed, and was updating (so Z-rip out, and A-rip in). | 19:03.04 |
| henrys | I see a few problems in the cet that don't look like practical issues, on the other hand I wasn't expecting your test to be "practical" | 19:04.03 |
| norbertj | henrys: how would it go with the regression tests. Run on your testfarm? or download the testfiles to our testfarm? | 19:04.22 |
| | would only be possible with (non)customer files I think. | 19:04.51 |
| henrys | I thought you had signed an NDA | 19:05.19 |
| norbertj | That I have, but non-the-less. Problem files from other customers clients ??? | 19:05.55 |
| Robin_Watts_ | norbertj: We'd give you access to our testfarm. | 19:06.06 |
| | (known as 'the cluster') | 19:06.22 |
| norbertj | robin_watts: would be ok with me.. | 19:06.40 |
| Robin_Watts_ | You'd be able to kick off jobs, and then review the results. You'd never actually need to get your hands dirty with other peoples files if you didn't want to. | 19:07.21 |
| henrys | norbertj:looking at the test results I think your change is progress and the font selected setting was just masking another issue. | 19:07.43 |
| | I'm going to put it in and create a new bug for the new issues. | 19:08.41 |
| tor8 | Robin_Watts_: did you update the news.html? if so, you've got the date wrong :) | 19:09.07 |
| norbertj | I think that would be ideal, especially with a problem/fix like this last one. Could give me early warning if there was a regression. And testing with our testfiles/farm I was planning to do next week, but would also take a couple of hours to complete. | 19:09.31 |
| tor8 | Robin_Watts_: not super thrilled about posting news on the index page, but if we want to do it to get attention for a while I'm fine with it | 19:09.36 |
| Robin_Watts_ | fixed. | 19:09.44 |
| | tor8: It seemed to fit nicely there. The old text said "We don't do interactive features like forms, javascript or transitions" | 19:10.17 |
| henrys | norbertj:how many cpu hours for your test? | 19:10.23 |
| tor8 | Robin_Watts_: right. it looks good to me, we can remove the boldness and rewrite it once it's out of preview. | 19:10.57 |
| Robin_Watts_ | After the next release... indeed. Boldness? | 19:11.06 |
| norbertj | 3-4 hours on a farm of I think 10 pc's / configuration. Yes, we have a lot of files. | 19:11.23 |
| Robin_Watts_ | Oh, the TECHNOLOGY PREVIEW ? right. | 19:11.28 |
| tor8 | Robin_Watts_: the <h2> boldness | 19:11.38 |
| Robin_Watts_ | right. | 19:11.43 |
| | I declare this release complete! | 19:12.46 |
| henrys | norbertj:ours just runs under 1/2 hour on about 10 pc's but we do more complete overnight testing | 19:12.59 |
| tor8 | Robin_Watts_: good job! | 19:13.20 |
| Robin_Watts_ | Now if sebras finds a showstopping bug, hit him :) | 19:13.42 |
| norbertj | We have a room completely stuffed with pc's (100+) that are only doing tests, whole day long. We are running about 10 products (b&w, color). | 19:13.47 |
| | henrys: The wife is calling, have to go to table (diner). | 19:15.35 |
| henrys | norbertj:well chrisl_away isn't here right now and he creates the cluster account we'll have him set you up and send you an email with instructions ttyl | 19:15.56 |
| | Robin_Watts_:looks good! At risk of an infinite loop how does miles news release reference the download? | 19:22.40 |
| Robin_Watts_ | Has Miles not released already? | 19:24.02 |
| | He was planning to get the release out by Oct 31st. | 19:24.18 |
| | Cos that was the deadline for the AndDevCon news releases. | 19:24.27 |
| | If he hasn't released yet, I'd just send people to mupdf.com | 19:24.47 |
| | cos it's obvious from there, I hope. | 19:24.55 |
| henrys | yes I imagine he has released. | 19:25.03 |
| | I was thinking he could do an addition with where to get it and all but probably not necessary. | 19:25.38 |
| Robin_Watts_ | I can see how to load the file into javascript. | 20:05.18 |
| | I can create canvases and draw onto them. | 20:05.27 |
| | It turns out that you can even capture the output of each canvas as a png, and hence not have to regenerate it multiple times. | 20:06.05 |
| | canvases even support compositing, so I could just put the pngs on top of one another and composite them together to get the final image... except there isn't a 'darker' blend mode. | 20:07.03 |
| | There is a 'lighter' blend mode, but no 'darker' one. WTF? | 20:07.17 |
| | Can I generate the inverse, then use lighter then invert it? | 20:08.25 |
| norbertj | Bye guys, it's weekend for me now. | 20:08.58 |
| Robin_Watts_ | night norbertj | 20:09.12 |
| henrys | bye norbertj | 20:09.15 |
| henrys | will proof read his commit message - 50x on the board | 20:53.22 |
| mvrhel | bbiaw | 21:34.21 |
| henrys | marcosw:all these multithreaded pcl rendering bugs - are there associate bugzilla reports for any of this? | 21:43.45 |
| marcosw | henrys: let me check | 21:44.23 |
| | There is this one: http://bugs.ghostscript.com/show_bug.cgi?id=693361 | 21:45.26 |
| | presumably many are showing as new errors\ because they are indeterministic (i.e. tests__pcl__thesis10.pcl.ppmraw.600.1 failed 7 out of the last 8 weeks but not last week). | 21:49.00 |
| henrys | thanks marcosw, maybe we should up the priority on those, I have an appointment bbiab | 21:53.12 |
| | marcosw:maybe I could make it P0 | 23:28.07 |
| | Forward 1 day (to 2012/11/10)>>> | |
IRC Logs