| <<<Back 1 day (to 2014/09/11) | 2014/09/12 |
kens | OK so I don't believe I can make pdfwrite capture a CIDFont with a BuildGlyph procedure (eg a missing Bold CIDFont for which we create a synthesised version). However, that doesn't produce a bitmap either... Since the glyph is shown as a fill and charpath stroke, we end up emitting the path lots of times though. | 10:38.26 |
chrisl | It's a stupid implementation anyway :-( | 10:41.59 |
kens | Yeah, but it shows up a class of fonts I simply can't capture this way. Which means I need to be able to differentiate between fonts I probably can capture, and fonts I shouldn't try to. | 10:42.56 |
| Time for a flag I think. | 10:43.10 |
chrisl | kens: on Windows, if you're doing a "current user only" install, is there a convention about where to stick the files (given that a "normal user" can't always write to Program Files)? | 10:53.16 |
kens | Not that I know of. | 10:53.29 |
| Possibly 'Users/<username>' ? | 10:53.47 |
chrisl | That's what I was thinking of, but figured there might be a preferred option of which I'm not aware..... | 10:54.28 |
kens | I'm no longer clear on where MS expects you to install stuff, I got the impression that installing stuff to Program Files was frowned on generally now. TBH I'd just try and stick it in Program Files, and if it doens't work then they are on their own | 10:55.02 |
chrisl | Yeh, the problem is, we'll just error out - I'm not sure whether I can backtrack to the install path dialogue | 10:55.56 |
kens | Erroring out sonds OK to me :-) | 10:56.12 |
chrisl | Okay, perhaps that's okay, if I can catch the error, and emit a legible message | 10:57.15 |
kens | I'm not sure how you're going to test this.... | 10:57.33 |
chrisl | Test what? | 10:58.17 |
kens | Failing to install like that | 10:58.30 |
chrisl | I have a win7 machine with a "normal" user setup (it may or may not still have UAC active!) and that normal user can't write to Program Files | 10:59.17 |
kens | Oh, OK that makes sense then. | 10:59.40 |
pedro_mac | chrisl: tbh, thatâs the intention of the user privs, so seems reasonable that install should fail. Nicer if we could warn before attempting to install | 11:00.10 |
chrisl | pedro_mac: I was hoping to have a different default install directory if we're running as a current user only install.... | 11:01.05 |
pedro_mac | chrisl: sure - makes sense to be in the user directory as kens said in that case | 11:02.08 |
chrisl | Now I just have to work out how to make that happen...... | 11:02.42 |
pedro_mac | GetUserProfileDirectory is probably a reasonable option | 11:05.29 |
| although it varies a bit in location based on the OS version | 11:05.44 |
chrisl | I need to fathom how to do it in nsis speak..... | 11:07.03 |
pedro_mac | ah :) | 11:07.13 |
| can you get at environemnt vars? | 11:07.25 |
chrisl | Yeh, but nsis also has "proper" ways to get at some stuff - it seems the user's home directory is called "PROFILE" in the nsis world.... | 11:08.15 |
pedro_mac | cool | 11:09.29 |
chrisl | Ugh, two Windows VMs really kill my machine! | 11:12.25 |
kens | OK so that should 'fix' CIDFont type 10 (CIDFont with BuildGlyph). I wonder what's the next thing that doesn't work..... | 11:12.37 |
| Mind you, the text code in pdfwrite has reached the point of complete incomprehensibility I think. I no longer understand what large parts of it do. | 11:13.37 |
chrisl | I think it passed that point *long* ago! | 11:14.16 |
kens | Oh no, it was just *mostly* incomprehensible before, now its total madness | 11:14.48 |
chrisl | Okay, so it's taken about 20 hours to copy just under 1Tb of data from my old NAS to my new one, over the network - I'm pretty impressed by that...... | 11:16.47 |
kens | chrisl, we seem to still have epswrite in the 9.15 release candidate :-( | 13:32.13 |
chrisl | Hmm, was I supposed to remove that? | 13:33.04 |
kens | Or maybe I was, it needs to go though | 13:33.11 |
| idiot users complaining on support and trying to use it | 13:33.38 |
chrisl | OKay, I'll zap it | 13:34.22 |
kens | THanks | 13:34.26 |
chrisl | kens: what about the scripts in lib - remove or just change them to eps2write? | 13:36.31 |
kens | I thought I had already changed them :-( | 13:36.41 |
chrisl | I still see eps2eps using epswrite | 13:37.15 |
kens | Yeah me too | 13:37.19 |
| Just change to use eps2write I'd say | 13:37.32 |
chrisl | And kill it in the makefiles, too? Or leave it for "safety" for a while? | 13:38.15 |
kens | I'd kill it in the makefiles too | 13:38.31 |
| Get rid of it completely or idiots like this will keep on using it. | 13:38.49 |
| Why on earth they htink a PDF->PS->PCL conversion is better than PDF->PCL escapes me completely | 13:39.18 |
chrisl | So is gdevps.c completely unused now? | 13:39.38 |
kens | I believe it is yes, which is a major goal of getting rid of epswrite | 13:39.54 |
chrisl | I'll remove it from the build, and see what breaks :-) | 13:40.12 |
kens | Yeah, then I can find out what I still need to get rid of :-( | 13:40.25 |
| Wel, the ps2write output is less than half the size of the psmono output, and looks way better when rendered with GS at screen resolution (no surprise). But the area covered is identical | 13:46.44 |
chrisl | But we had another customer wittering about margins with the PCL output - possibly something related? | 13:47.39 |
kens | No, its the same customer :-) | 13:47.48 |
chrisl | Really? I'm astounded they have the brain to write their own device! | 13:48.15 |
kens | I don't think htye do, its a modification of an existing device I believe. And its *ancient* circa 1995, so I'm fairly sure this clown didn't write it | 13:48.48 |
| In fact the Author is given as Jim Walker | 13:49.15 |
| 'based uopn gdevdjet.c' | 13:49.33 |
chrisl | is still surprised...... | 13:50.16 |
kens | Intriguingly, the ljet4 output for the ps2write PostScript is 2x the size of the psmono output. I guess it doesn't compress 24-bit colour as well as it does monochrome. I didn't bother to set the colorconversionstrategy | 13:51.12 |
chrisl | I suspect it's a simple RLE compression, so the psmono output will have lots of nice runs to compress | 13:52.12 |
kens | Yeah that's what I was thinking | 13:52.21 |
chrisl | The cluster doesn't test epswrite does it? | 13:52.51 |
kens | No | 13:53.00 |
chrisl | I'll do a push anyway - I have time while my Windows test build completes | 13:54.50 |
kens | OK | 13:54.56 |
| THere does seem to be a difference in the ps2write and psmono output, I suspect that ps2write doesn't request a media size | 13:55.27 |
| So they could use PSDocOptions to get round that | 13:55.59 |
chrisl | Really? I thought it did...... | 13:56.05 |
kens | Well, I'm not certain | 13:56.17 |
| Onl;y just got as far as looking at the output in PCL | 13:56.31 |
| Of course, why they are going via PostScript in the first place eludes me, other than 'you are a fool sir' | 13:57.02 |
| The ps2write output only executes setpagedevice if you select -dSetPageSize :-) | 13:59.01 |
| Oh, but that is true, strange | 13:59.22 |
| Ah, it set the page size to 595x842, not 612x792 | 14:01.28 |
| LOL well, if I use 9.05 for testing (which I was doing only to get the psmono device for comparison) then the ps2write output comes out on A4 (595x842). If I use HEAD, the ps2write comes out on 612x792..... | 14:08.09 |
| Still doesn't render the same after conversion to PCL though | 14:09.43 |
chrisl | kens: still waiting on the Windows build to complete, but in the meantime: http://git.ghostscript.com/?p=user/chrisl/ghostpdl.git;a=commitdiff;h=94786ac0 | 14:16.05 |
kens | Crumbs, that was more extensive than I anticipated | 14:16.25 |
chrisl | Most of it is historical stuff that we don't actively support now | 14:16.53 |
kens | Yeah I see that | 14:17.00 |
| Looks fine to me | 14:17.13 |
chrisl | I think the most important part is: "gs/devices/vector/gdevps.c [deleted file] patch | blob | blame | history" | 14:17.49 |
| :-) | 14:17.57 |
kens | Yes, I noticed that :-) | 14:17.59 |
chrisl | And possibly: | 14:18.18 |
| -epswrite.dev: ken | 14:18.20 |
| +epswrite.dev: obsolete | 14:18.26 |
kens | Even better! | 14:18.35 |
| Hmm, so it does seem there is a problem with ps2write, its not permitting the page size to be set from the MediaBox, even when I select -dSetPageSzie | 14:20.11 |
| Good grief, now another one bleating. If they insist on using PCL as an input, they aren't gooing to get good results.. And a 25 page docuemnt to demonstrate it :-( | 14:26.36 |
pedro_mac | kens: just a typical Friday afternoon scenario⦠| 14:28.24 |
kens | I could use another holiday...... | 14:28.38 |
pedro_mac | :) | 14:28.43 |
kens | I'll let henry reduce that one to a sensible size first. | 14:29.05 |
henrys | will do kens | 14:32.30 |
kens | Ah didn't realise you were there, thanks henrys | 14:32.41 |
chrisl | kens: are you (and your phone) free? | 14:32.56 |
kens | suspects it'll be something weird about the way those particular barcodes are drawn | 14:33.05 |
| chrisl, let me check, Stella was phoning someone | 14:33.18 |
| chrisl, no not yet. | 14:33.45 |
| I'll call you when the phone is free | 14:33.52 |
chrisl | kens: okay - I have to finish early today..... | 14:34.07 |
kens | Yeah I guessed that | 14:34.12 |
| Not a problem, I'm feeling tired already :-( | 14:34.21 |
mvrhel_laptop | morning | 15:18.17 |
Robin_Watts | morning | 15:22.46 |
henrys | kens: did you want to work on this pcl stuff now or can it wait for marcosw1 to process it and create a bug ... (Monday) | 16:10.34 |
| kens: well monday for you. | 16:10.52 |
kens | henrys it can wait as long as you like :-) | 16:10.54 |
| I'm still wrestinglwith fallback fonts as outlines | 16:11.09 |
mvrhel_laptop | henrys: making good progress on the high level image stuff for the xpswrite device. | 19:40.39 |
| got the icc profile getting packed in there for the image. now getting ready to add in the tiff image | 19:41.02 |
| then I will add in the markup to use the proper brush | 19:41.17 |
| heading out for a bit now | 19:41.50 |
Atom2 | anybody here? seems to be pretty quiet ... | 22:50.47 |
rayjj | Atom2: it's pretty late on a Friday -- the European team is probably gone. Some of us on the West Coast of the US are still around | 23:49.58 |
| (it's probably been about an hour since Atom2 asked) | 23:50.39 |
Atom2 | rayjj: I am still here - and btw also in Europe ... | 23:57.32 |
rayjj | Atom2: working late :-) | 23:58.08 |
| what's your question/issue ? | 23:58.21 |
Atom2 | yeah, and facing a problem that I am trying to solve: Currently up to installing ghostscript on BSD as a dependency and I am failing misearbly | 23:58.48 |
| the compile process bails out with errors about undefined references in libgs.so having to do with gdev_vector_ stuff | 23:59.36 |
| Forward 1 day (to 2014/09/13)>>> | |