| | <<<Back 1 day (to 2015/11/26) | 20151127 |
| rsc_ | chrisl: My postscript knowledge is very bad. | 14:48.08 |
| rsc | The document in question is generated anyway. | 14:48.42 |
| kens | Hmm, did I miss a queston there ? | 14:49.04 |
| rsc | Don't know, was: [12:17:35] < rsc_> Any pointer how to debug output like this? http://fpaste.org/295101/48623015/ | 14:49.21 |
| kens | Well fundamentally, you don't, unless you are an experienced PostScript programmer. | 14:49.51 |
| chrisl | rsc: and even an experienced Postscripter isn't going to get anywhere without the source file! | 14:50.15 |
| kens | That looks like binary is being interpreted as tokens, which will usually happen if you have done something like transfer the vfile via a 7-bit medium (eg email) or done CR/LF translation on it | 14:50.28 |
| | I think both I and ghostbot were dicsonnected from IRC at the mie this was asked. | 14:51.10 |
| | SO I didn't see it, and there was no log either :-( | 14:51.27 |
| rsc | In theory, there is no translation. | 14:51.32 |
| kens | I have no idea what you mean by that | 14:51.44 |
| | As chrisl says, nobody can help you at all without seeing the source PostScript program | 14:52.16 |
| | Of course you could always try using a more recent version of Ghostscript, 9.07 is 2 years old | 14:53.20 |
| rsc | The program generating the postscript or the postscript file? | 14:53.30 |
| | Well, that is RHEL 7 including all updates. Interestingly it works with ghostscript from RHEL 6 - which is older. | 14:53.48 |
| chrisl | A Postscript file *is* a Postscript program | 14:53.58 |
| kens | The #PostScript file', PostScript is a programming language, so the PostScript is a program | 14:53.59 |
| rsc | Ah. | 14:54.22 |
| chrisl | You will have to find a way to grab the Postscript file, as a first step | 14:54.52 |
| rsc | Yes, that I have. | 14:54.59 |
| kens | Then make it available somewhere | 14:55.12 |
| | The fact that RHEL is shippign a given version in their package doesn't mean its the latest version, lots of distributions are very slow about updating packsages. The version number you quote (in the pastebin) is 2 years old. Its entirely possible there's a bug fix for that problem. | 14:59.08 |
| rsc | May I send the URL to the file via /msg? The file shouldn't go public ;) | 15:00.02 |
| kens | Yes that will be OK | 15:00.13 |
| | best send it to chrisl *and* me | 15:00.19 |
| chrisl | RHEL is always a ways behind Fedora core - it's <ahem> more stable....... | 15:00.41 |
| rsc | Yes ;) | 15:01.06 |
| chrisl | The problem is, you avoid new bugs, but you are saddled with old ones | 15:01.38 |
| rsc | Fortunately none of my "postscript" problems so far was really a postscript problem :) | 15:02.01 |
| kens | Well it still fails on the 9.18 release | 15:02.49 |
| chrisl | Hm, I didn't get the URL..... | 15:03.25 |
| kens | I'll send you the file | 15:03.41 |
| rsc | chrisl: sorry, now you have. | 15:03.45 |
| kens | Or that works | 15:03.54 |
| chrisl | Thanks got it! | 15:03.55 |
| kens | Thre are two lumps of binary data, and image and a font | 15:04.04 |
| rsc | The image is likely the cause. | 15:04.13 |
| kens | I thnk the font is more likely | 15:04.20 |
| | The image is ASCII85 encodede | 15:04.52 |
| | So you wouldn't get binary like you saw | 15:05.00 |
| chrisl | Hmm, different error...... | 15:05.58 |
| kens | The program uses, but does not include, DejaVuSans | 15:06.08 |
| | I get an undefined error | 15:06.12 |
| chrisl | I'm getting a typecheck | 15:06.26 |
| | But that is using the Ubuntu installed gs | 15:06.45 |
| kens | actually that's not a font I see there, I was misled by the usage of DejaVuLGCSans | 15:06.48 |
| rsc | I have a similar document (actually the same) but generated on RHEL 6, wait. | 15:07.13 |
| | And that one works. Whyever. But maybe helps. | 15:07.23 |
| kens | I'm getting undefined on both 9.18 and current code | 15:07.23 |
| | Well only in as much as it may tell you that the problem is in the generating applicatoin..... | 15:07.53 |
| chrisl | Yes, current code, me too | 15:08.10 |
| kens | I thnk the file is broken | 15:08.29 |
| | THere's a 'PBEGIN' which looks like its intended to include an EPS, but it actually ioncludes a JFIF file | 15:08.51 |
| | You can't just stick a JFIF into a PostScript program, it needs support | 15:09.13 |
| | SO I believe that's the problem, your PostScript file is not valid | 15:09.49 |
| chrisl | Yeh, the first PBEGIN does indeed prefix a "real" EPS | 15:10.28 |
| | The second binary | 15:10.44 |
| kens | But hte second one doesn't, it just dives straight into the JFIF file | 15:10.49 |
| | Your working file has an EPS where the non-working file includes JFIF data | 15:11.37 |
| | I would have to say that the generating application is at fault. | 15:11.55 |
| chrisl | Indeed. If I remove the JFIF binary, the file runs without error | 15:12.21 |
| kens | If I remove the JFIF data from the first file it works | 15:12.24 |
| | echo :-) | 15:12.31 |
| chrisl | Well, there's your problem! | 15:12.42 |
| kens | The missing fonts mean it looks *awful* mind | 15:12.48 |
| chrisl | Worse, missing *CIDFonts*.... | 15:13.24 |
| kens | hadn't looked that closely | 15:13.34 |
| chrisl | "Can't find (or can't open) font file %rom%Resource/Font/DejaVuLGCSans-Identity-H." | 15:13.50 |
| kens | I htnk it uses regular DejaVuSans as well | 15:14.11 |
| | Good Grief, its creates a temporary dictionary called DejaVuSans | 15:14.50 |
| | Oh I see, its defining a type 3 font, ick | 15:15.03 |
| | And it doesn't even make a font out of it, what a pile of poo | 15:16.21 |
| | In any event, the bottom line is that the application you are using generated a broken PostScript program, nothing we can do about it. | 15:17.31 |
| | The only reference I can find to mercaware suggests its a commercial ERP application | 15:18.36 |
| | I should report it to them as a bug | 15:18.57 |
| chrisl | In the very strongest possible terms! | 15:19.41 |
| kens | I've just deleted the PostScript programs | 15:20.48 |
| chrisl | rsc: do you have all the information you need? | 15:21.38 |
| rsc | Okay, so the issue is that JFIF data is put into the postscript directly rather EPS? | 15:23.03 |
| kens | Braodly | 15:23.33 |
| | Broadly | 15:23.37 |
| | You can include a JPEG in PostScript, but you have to tell the program that its an image, and its DCT encoded. Ths program does neither | 15:24.06 |
| rsc | The CIDFonts issue is already known, but not easy fixable :( | 15:24.08 |
| kens | Well, I'd say that's wrong too. CIDFonts should *always* be embedded | 15:24.32 |
| | And the usage of DejaVuSans is diabolical too | 15:24.47 |
| | Emitting the glyphs as linework | 15:24.56 |
| | THis is slow (no caching of glyphs) and results in poor quality as all the spcial care taken with fonts is not applied to basic linework | 15:25.25 |
| rsc | Unfortunately, I have to cover the intermediate time until the whole thing has been changed, thus I need to figure out why it puts a JFIF binary into it. | 15:26.36 |
| kens | Not a clue, that seems to me like a question for the mercaware developers | 15:27.00 |
| rsc | Yepp, indeed. | 15:27.18 |
| chrisl | rsc: I notice the EPS (or one of them) is created by Scribus - are you supplying the file to be embedded? | 15:28.27 |
| | The glyphs as lineart seems to come from Scribus, btw, which is a little worrying for a "professional" level DTP package..... | 15:29.57 |
| kens | :-( | 15:31.27 |
| rsc | chrisl: I thought the issue is JFIF? | 15:32.32 |
| | chrisl: the file is a mixture out of generated code, a background (EPS) and a image (JFIF) | 15:32.50 |
| chrisl | rsc: It is. I'm just wondering if you're giving a JPEG file to embed, rather than an EPS | 15:33.11 |
| | And the generating app isn't validating that it really is a valid EPS | 15:33.28 |
| kens | THe *error* is hte JFIF, we're just commenting on the quality of the remaining parts of the program | 15:33.31 |
| rsc | I give a JPEG file, but on RHEL 6 it seems to convert (?) it while on RHEL 7, it doesn't seem to convert it. | 15:33.46 |
| chrisl | rsc: well, as an interim solution you could convert the JPEG to EPS and try using that instead | 15:34.48 |
| | Emitting the font as lineart may be a setting in Scribus. I don't know it well enough to know | 15:35.58 |
| | Oh, ick, the EPS header says "PS-Adobe-3.0" and then it uses "colorimage" for the image | 15:42.29 |
| kens | Well the 3.0 just defines the DSC convention level, not the PostScript level | 15:44.21 |
| | Though I'm not convinced it is truly DSC-compliant | 15:45.07 |
| chrisl | Oh yes - I thought that was the "EPSF-3.0" bit.... | 15:45.12 |
| kens | I thnk that's means its DSC 3 compliant *and* an EPSF | 15:45.41 |
| chrisl | It's not like colorimage is wrong, even in LL3, it's just not very nice..... | 15:46.00 |
| kens | True | 15:46.11 |
| | But the whole PS program is not very nice :-) | 15:46.22 |
| chrisl | To be fair, I've seen much, much worse! | 15:47.42 |
| | I mostly prefer this to Adobe's PS output | 15:48.10 |
| kens | Well, its readable, but the missing CIDFont and the JFIF binary are pretty bad | 15:48.44 |
| chrisl | I was looking at the earlier file, so without the JFIF issue. I'm surprised that the CIDFont issue is considered "not easy" | 15:50.34 |
| kens | Well I've deleted all the files now so I can't really comment.... | 15:51.03 |
| rsc | kens, chrisl: Thank you very much so far! | 16:07.04 |
| | Forward 1 day (to 2015/11/28)>>> | |
IRC Logs