| <<<Back 1 day (to 2019/09/23) | Fwd 1 day (to 2019/09/25) >>> | 20190924 |
rnissl | Hi kens, I've got a further regression with 9.28rc3. | 14:10.22 |
| gswin32c" -dBATCH -dNOPAUSE -q -c .setsafe pstack | 14:10.46 |
| GPL Ghostscript RELEASE CANDIDATE 3 9.28: Unrecoverable error, exit code 255 | 14:10.46 |
kens | Hmm, let me pass you to someone else, I'm up to my ears right now. chrisl ping ? | 14:11.21 |
rnissl | -c .setsafe fails without -dNOSAFER | 14:11.26 |
kens | Well you don't need .etsafe, since we are already safe in this case. But I guess it shoudl not throw an error | 14:11.50 |
| OK seems chrisl is not about, I can confirm that calling .setsafe exits the interactive prompt, which it shouldn't do | 14:14.01 |
rnissl | I totally agree with you. In 9.28 -c .setsafe is not necessary as it is the default, and it should not throw an error | 14:14.05 |
kens | I'll try and find out why | 14:14.14 |
| BTW, thanks for finding it | 14:14.48 |
| testing is much appreaciated | 14:14.58 |
rnissl | well, I just wanted to start early this time -- last time I couldn't use official 9.27 due to some issues | 14:16.01 |
kens | We suspected there would be trouble this time, that's why we started the release so early | 14:16.23 |
| Looks to me like .setsafe probably should check the current SAFER and exit without doing anything if its set. I htink its trying to change stuff its not allowed to | 14:17.51 |
| Yes, its tryign to lock the file accesses and since tehy are already locked, that throws an error | 14:19.02 |
chrisl | Here now | 14:25.44 |
kens | wb | 14:26.24 |
| If you run GS and then execute .setsafe, it exeits with an error | 14:26.40 |
| unless you started with NOSAFER | 14:26.48 |
chrisl | Well, don't do that | 14:26.51 |
kens | Yeah but it really shouldn't exit I'd have thought ? | 14:27.02 |
| TBH I'm not exactly sure where its exiting, I'm not hugely familiar with all this new code | 14:27.21 |
| oh probably .addcontrolpath | 14:27.43 |
chrisl | Yes, probably | 14:27.54 |
kens | Yes, calling .addcontrolpath when we're already in safer mode calls gs_is_path_control_active, which returns true and we throw a fatal error | 14:29.23 |
chrisl | I suppose I can tweak .setsafe, but I'm rather surprised it didn't throw an error previously | 14:29.42 |
kens | I tried checking SAFETY /setsafe get in .setsafe and its not set | 14:30.18 |
| mereting time, lets adjourn for 30 minutes | 14:30.41 |
chrisl | We don't set SAFETY any more | 14:30.48 |
rnissl | gs-9.26 doesn't exit if .setsafe is called multiple times | 14:30.53 |
kens | chrisl I didn't know any other way to test if we had locked safety | 14:31.12 |
| And we can't (apparently) use stopped to get out of a fatal error | 14:31.24 |
| chrisl could we add a PostScritp operator for gs_is_path_control_active() and use that to not try and set path control to active (with a warning, I would suggest) | 14:35.11 |
chrisl | No, we didn't | 14:35.24 |
kens | I know we didn't, but would that be a reasonable solution ? | 14:35.48 |
chrisl | I'm looking at it now | 14:35.58 |
kens | Then I'll shut up | 14:36.05 |
chrisl | kens: https://git.ghostscript.com/?p=user/chrisl/ghostpdl.git;a=commitdiff;h=13344a38b1d0f9b882ffa3bb429603fec8237ac2 | 14:56.10 |
kens | chrisl I'd be inclined to emit a warnign if we execute .setsafe and we already haev the file paths locked. butotherise, yes that's exactly what I was thinking of | 14:57.39 |
chrisl | kens: I'm struggling with a wording that will mean anything sensible to an end user.... | 15:00.19 |
kens | Something like 'PostScript program executed .setsafe while already in SAFER mode' | 15:00.41 |
| Only savvy people should be using .setsafe (because it implies DELAYSAFER, or at least *some* knowledge of SAFER) | 15:01.11 |
Robin_Watts | chrisl: Did you have an opinion on the steak/steak/fish vs steak/chicken/fish question? | 15:01.33 |
kens | So ordinary users shouldn't see it, if they do it means their PostScript program was up to something suspicious | 15:01.37 |
Robin_Watts | Balls, wrong group. | 15:01.41 |
kens | chrisl its not so much end users I'm thinking of, but people like rnissl who might quite like to know that this has happened. | 15:02.25 |
rnissl | I do ;-) | 15:02.49 |
chrisl | kens: Yeh, I've been slightly trying to divorce "SAFER" from "file access controls" | 15:02.58 |
kens | Then say maybe 'while file access was already locked' | 15:03.13 |
chrisl | Which might confuse people who've called .setsafe..... | 15:03.38 |
kens | 'Attempt to lock file access, when file acces had previousl already been locked. Suspicious behaviour has been prevented to protect you' | 15:03.49 |
| for your protection* | 15:04.02 |
Robin_Watts | mvrhel_laptop: come to #artifex... you know you want to... | 15:07.26 |
chrisl | kens: https://git.ghostscript.com/?p=user/chrisl/ghostpdl.git;a=commitdiff;h=05e794d14a2853c45a08d5c49d50d1fa5cadaf6c | 15:08.47 |
kens | Looks fine to me | 15:09.10 |
chrisl | kens: I also have: https://git.ghostscript.com/?p=user/chrisl/ghostpdl.git;a=commitdiff;h=5ca729649efc3b041237b8b45bd2d011b70143ff | 15:09.24 |
kens | Umm, maybe you should get Robin to look at that one ? | 15:10.05 |
| I mean, it looks fine, but I'm not really up to speed with all that | 15:10.27 |
chrisl | At the moment if you give tiffsep(1) an output file like out.tiff, you end up with out.tiff(Sepname).tif for the separation files | 15:11.21 |
kens | Ah, I see, well that's certainly an improvement then | 15:11.44 |
chrisl | I found it confusing because the composite preview was still out.tiff | 15:12.17 |
kens | That would indeed confuse *me* ! | 15:12.34 |
chrisl | Okay, I haven't been able to induce a problem with Robin's and my access control changes, so I'm going to push all that stuff - assuming this cluster run is clean | 15:14.17 |
kens | Great, needs to be done. Planning to wait until next week for RC4 ? | 15:14.51 |
chrisl | Yes, probably - on the off chance someone else on the team decides to do some testing..... | 15:15.43 |
kens | And time for replies from the distros | 15:15.59 |
chrisl | And also, after next week, nothing can really happen for nearly two weeks | 15:16.34 |
kens | Yeah. | 15:16.45 |
kens | hopes we're done now..... | 15:17.02 |
chrisl | Well, I'm also thinking, maybe do another release test run next week | 15:17.33 |
kens | Ugh, well I suppose there have been a few changes. Presumably test against the RC1 tagged code ? | 15:18.09 |
chrisl | RC2, that was the last set of release tests we did | 15:18.39 |
kens | Oh OK, my mistake | 15:18.48 |
chrisl | There's really only two commits that *might* influence rendering (and one of those is unlikely at worst). | 15:20.38 |
kens | So it should be a quick and easy test :-) | 15:20.55 |
chrisl | Truly, I've lost count of how often I've been wrong about that! | 15:21.45 |
kens | I knew I was tempting fate even saying it | 15:22.01 |
chrisl | rnissl: Thanks for spotting the .setsafe issue | 15:22.48 |
rnissl | chrisl, you're welcome | 15:23.02 |
| ghostscript is and has been one of my favourite tools since the early nineties ;-) | 15:24.46 |
chrisl | rnissl: I just hope we're not putting you off! Unfortunately, the way things have gone has meant rethinking decisions made a long time ago, and the result is a certain level of upheaval for some users - for which we are sorry, but couldn't see an alternative | 15:27.45 |
ray_laptop | rnissl: sounds like you started with gs about the same time as me (version 2.5.1) | 15:35.11 |
| I don't recall the exact date that was released -- I still have 3.33 laying around and that was Oct 1994 | 15:36.08 |
| oh, google tells me 2.5.1 was 9/11/92 | 15:37.56 |
| version 1.0 was 8/11/88 | 15:41.17 |
| I should see if I can put together a git repository that goes way back (just for posterity) The Artifex git doesn't start until 1998 | 15:46.00 |
rnissl | chrisl, I don't mind the issues I had with rc1 to rc3. I just wanted to start testing erlier that with 9.27. | 16:02.17 |
chrisl | rnissl: That's good, and it is appreciated | 16:03.22 |
rnissl | ray_laptop, I think I have even used 2.4.x on my Amiga 3000T -- spent quite a while to compile it ;-) | 16:04.21 |
| and around 1995 I reported a couple of issues with tiff g4 encoding to L. Peter Deutsch | 16:06.34 |
| will leave for today -- thanks for your support, looking forward to rc4. | 16:28.55 |
kens | bb and thanks again for the testiong | 16:29.14 |
| <<<Back 1 day (to 2019/09/23) | Forward 1 day (to 2019/09/25)>>> | |