IRC Logs

Log of #ghostscript at

 <<<Back 1 day (to 2015/05/07)20150508 
kens Well, a desperate idiot Indian. mail to support, bug report and a posting on SO :-)07:55.06 
chrisl Clearly so. If he posts an actual test file, I can try it on SPARC07:56.35 
kens I thnk I pointed that out in the bug report07:56.51 
  It would be interesting to test on a Spar07:57.03 
  Grrr Sparc...07:57.09 
chrisl You did, yes. SPARC is the only platform I have that enforces alignment, and the "bus error" rather suggests that might be the problem07:57.44 
kens Yup, that's pretty much what I thought. On SO I told him he was going to have to do some debugging (it is a site for programmers after all :-)07:58.25 
chrisl Even if his job description suggests developer, I'm sure he's not going to be up to that07:59.58 
kens Nope :-)08:00.15 
sebras kens: chrisl: doesn't a suitably configured ARM also enforce alignment?08:33.28 
  i.e. pi or something like it.08:33.42 
kens sebras : possibly08:46.29 
  BUt to be honest, its not something I'm going to look into for a free user who hasn't done any work himself08:46.56 
  Oh, and so that a global logistics company can use GS for free.....08:47.16 
Robin_Watts sebras: Some ARMs enforce alignment. Some ARMs just rotate the results. Some ARMs abort.08:55.55 
kens \Well, my Pi has a ARMv7l, let me see...08:57.22 
  Apparently it raises an abort if the source/destination address is not word aligned08:57.58 
  Oh, though it also has a compatibility mode for non armv7 software08:58.45 
  The RISCOS page says that gcc 4.1.1 produces compatible code.08:59.33 
  Anyway, still not going to look at it.09:02.14 
chrisl My RasPi didn't show a problem when I tried it for a similar issue before - I'd assumed it was a hardware/boot config option (like endianness)09:17.55 
kens I'm seeing conflicting info on various pages09:18.13 
  So it may be configurable, I can't find a proper data sheet anywhere09:18.30 
Robin_Watts kens: On some ARMs it is certainly software configurable.09:18.56 
chrisl I didn't pursue it at the time, though, as I just had to find a power chord for my SPARC to work on the actual reported platform09:18.58 
Robin_Watts suspects you mean a power cord. Though "chord" does conjure up a much better image :)09:19.39 
kens Robin_Watts : some thngs ce3rtainly seem to be configurable on the v7 but I couldn't (quickly) find a dtaa sheet so I gave up09:19.57 
chrisl Grr, just been talking guitars with someone else.....09:20.08 
kens So hrisl is playing rock on his sparc ?09:20.11 
chrisl TBH, it might be better as a musical instrument than as a computer these days!09:20.33 
kens can understand the lack of a power chord on a Sparc these days09:20.49 
  Given that Chris can test alignment (fairly) easily, I gave up looking for more info on ARM alignment09:21.34 
Robin_Watts kens: It's configuration typically done by the kernel.It's not something you can generally switch easily to do tests. So I agree that you shouldn't chase it.09:24.15 
kens Yeah I was just reading the Linux how to on the subject09:24.31 
  It does look like Linux should throw an alignment error, whch is trapped in the kernel and I *think* should give a seg fault.09:25.02 
  Sorry ARM should throw an alignement error.....09:25.15 
  There's infor on finding out how its set up, but hey, getting bored now09:25.50 
chrisl You can set it to seg fault, too.09:26.01 
  But also if only enforces alignment to word boundaries, that's not the same as SPARC - I can't remember what IA64 does....09:27.03 
kens Beats me, I'd have to go look that up too09:27.20 
  TBH its up to the person askign to reproduce it on a more mainstream device, or debug it himself.09:28.01 
  He obviously must have *built* it himself.....09:28.12 
chrisl I also don't want to waste any more time on (probably) a freeloader09:28.26 
kens Oh I'm sure they are a freeloader. I bet they want their reports (invoices etc) in PDF format, and tehy currently come out in PCL, so they want to use GS to do the work.09:29.05 
tor8 Robin_Watts: Thanks, and yes, I think we'll need someone to build and upload the iOS version.09:36.09 
  sebras: font size can be changed with a command line argument09:36.28 
  sebras: I need to add @page at-rule parsing and handling for customizable margins, then we can add user stylesheets09:37.02 
sebras tor8: ah, I didn't notice that.12:03.09 
dyfrgi x86-64 enforces alignment for some instructions. There are non-enforcing instructions for the same thing, but the alignment-enforcing ones are used by gcc sometimes.12:55.37 
kens I'm not sure NOrbert is listening to me.....13:29.41 
henrys kens: yeah that the one where the raster op device is loaded in the image code and pdfwrite never sees it.14:26.53 
kens henrys, I guessed it was a ROP14:29.36 
henrys kens: oh lord hin-tak is on it.14:29.53 
kens O.O :-(14:30.02 
  Well he's more or less saying pretty much what I already did. Its not part of the graphics model, in the general case we can't handle it in PDF, so an image is all we can do.14:31.10 
  I don't think the bug he's referring to is relevant either, since the output is to pdfwrite.14:32.18 
henrys kens: yes he's just bringing up stuff stuck in his craw14:34.00 
kens Not helpful......14:34.23 
chrisl If we make a bug private, can the original reporter still see/post on it?14:35.25 
kens I don't know what (if anything) you thnk we should do about this Henry ? It may be that ths particular ROP cna be transformed into something acceptable to PDF< but we know that in general ROPs can't.14:35.45 
henrys kens: I'm going to close it.14:36.18 
kens Fine by me :-)14:36.26 
rayjj darn, yet another 'membership disabled' message from gs-devel (and with the same confirmation string). looking at my trash, the first time this message was sent was on Apr 23rd, and I probably missed it.14:41.15 
chrisl rayjj: if you want to try shouting at GMail, then have it - we rather figured they'd just ignore us.....14:42.20 
henrys kens yeah you analyzed this same issue 3 years ago: comment #3 . Don't you remeber ;-)14:43.24 
kens I have the same kind of early Alzheimer's as Robin.......14:43.47 
  Chris on the other hand seems to have the proverbial elephant's memory14:44.04 
chrisl For some things.....14:44.20 
Robin_Watts kens: I have what?14:44.39 
kens Eh what ?14:44.47 
henrys how are ya'll doin' with the new gov't?14:45.15 
kens Its less amusing than it could have been :-)14:45.30 
Robin_Watts henrys: It's like the old government, but with the stabilisers taken off.14:46.12 
henrys I just read the conservatives did a bit better14:46.44 
chrisl I can't say I'm surprised by the LibDems' results....14:46.47 
Robin_Watts In 18 months we'll have an in/out referendum for europe.14:47.00 
  I suspect at that point, Scotland will want to stay in, and the rest of the uk will want out.14:47.23 
henrys you guys should be a separate continent14:47.31 
Robin_Watts and that'll set us up for another independence question.14:47.41 
  henrys: We've had continents before, but you and India wanted out :)14:48.23 
henrys you can barely hold on to scotland now... 14:48.53 
kens I think Millipede's comments about the SNP probably didn't help Labour14:51.30 
Robin_Watts s/'s comments about the SNP//14:51.48 
kens Well that too, but I meant in Scotland14:52.06 
  TYring to threaten people, and coming over all strong man when Sturgeon was being reasonable doens't really play well14:52.32 
Robin_Watts I think the SNP massively helped the conservatives :)14:52.34 
kens Yeah, Cameron must be delighted14:52.47 
Robin_Watts "On friday, I'll be sitting across a table from Ed Milliband discussing how to kick the Tories out".14:53.02 
kens The referendum was a bit of a gamble, but its paid off for him big time14:53.07 
Robin_Watts That played really well in england.14:53.31 
kens No doubt14:53.38 
  Same as the '10% holding to ransom' played well in Scotland14:53.52 
Robin_Watts Gotta reboot. Windows is evidently a labour supporter and sulking...14:54.18 
  .quit Pop!14:54.22 
chrisl And yet again, our gs-regression mails are being filtered as spam - HOW MANY TIMES GMAIL, HOW MANY TIMES? :-(14:56.54 
kens Yeah, mine too, but only some of them......14:57.06 
  And you have to individually forward them too (or if you can mass forward I can't figure out how)14:57.34 
chrisl I just copy them14:58.05 
kens CC ? How does htat help ?14:58.28 
chrisl No, I copy them out the spam folder, into my regression folder14:58.45 
kens Ah but I want them sent to another account14:58.54 
  Mail whch doesn't get spam filtereed is auto-forwarded to another account (because I cna't do password authenticaton on commercial GMail). But if it goes into spam, I have to mark it as not spam and forward it manually14:59.43 
Robin_Watts chrisl: I suspect bug 695913 should be thee, not me.16:15.33 
  I can try to verify it if you want.16:15.48 
chrisl Okay, I'll take it16:16.48 
henrys huh the internal font name encoded in the font for the urw wingdings we got recently is URW Dingbats. I wonder if that is intentional to avoid trademark issues or a mistake. It is confusing since Dingbats is a different font.16:32.25 
  my turn for forgetfulness, I already knew that.16:41.58 
  marcosw: for the logs I've sent off the ghostpdl support issue to scott and miles so they can try and get a support contract.16:57.30 
Robin_Watts I see the problem with the bicubic scaling.18:33.09 
  bicubic scaling is suitable for *upscaling* only.18:33.20 
  For downscaling it just doesn't work. It's actually worse than linear.18:33.56 
 Forward 1 day (to 2015/05/09)>>>