On 2017-02-01 10:26 PM (-4), Cousin Ricky wrote:
> Those first two scenes did not actually use the data in the 1 megabyte
> file, so I dropped it, concatenated everything else into a single file,
> then took a hatchet to the file.

Correction: only one of them didn't use the 1 MB file.  I'm not thinking 
straight today.

Post a reply to this message

Am 02.02.2017 um 03:26 schrieb Cousin Ricky:
> On 02/01/2017 07:49 PM, Cousin Ricky wrote:
>> On 2017-01-11 11:37 AM (-4), clipka wrote:
>>> If that fails, please provide us with the corresponding scene file, so
>>> we can try to reproduce the problem on a Windows version, where we have
>>> more debugging aids at our disposal.
>>
>> The very first two scenes I tried failed in the same manner that
>> LanuHum's scene did.  (They also failed with alpha.8927145.)  Since
>> then, all of the scenes I've tried have rendered successfully.
>>
>> Unfortunately, both of those first two scenes use an include file that
>> is more than 1 megabyte.
> 
> Those first two scenes did not actually use the data in the 1 megabyte
> file, so I dropped it, concatenated everything else into a single file,
> then took a hatchet to the file.
> 
> Guess what?  The one thing those two scenes had in common that all my
> other tests did not was Greek text.
> 
> This appears to be the same charset utf8 bug that ThH uncovered.  And
> since LanuHum probably uses a lot of Cyrillic text, I'm guessing that
> his is the same problem.

Sounds totally plausible.

If that is the case, then the next beta should fix it.

Post a reply to this message

Am 01.02.2017 um 23:30 schrieb Cousin Ricky:
> On 2017-01-11 01:36 PM (-4), clipka wrote:
>> It would be interesting to know whether the latest tagged alpha
>> (v3.7.1-alpha.8927145) works better than the beta. Since we've made only
>> very few changes to the source code since then, that would give us a
>> clear hint where to search for things that might be going wrong.
> 
> Alpha.8927145 crashes.  Alpha.8917471 runs splendidly.

That fits well with the theory that it's the "utf8" bug you're experiencing.

Post a reply to this message

Am 02.02.2017 um 00:38 schrieb Cousin Ricky:

> The parser chokes on .ini files that are not in the current directory,
> but are in the library path.
> 
> I wish I could say in what manner it choked, but the message was that
> infuriatingly unspecific:
> 
>   Problem with option setting
>   [Command line is echoed here.]
>   Failed to parse command-line option
> 
> Once again, if the parser had given a more helpful message, I could have
> saved nearly an hour and a half figuring out what went wrong.  I've been
> complaining about this for five or six years now.  PLEASE FIX THIS!

This seems to be one of various shortcomings of my arch nemesis, the
"VFE" layer, which appears to provide no way whatsoever to pass on any
details why it rejects the command-line.

I've been gaining ground in my uphill battle against the VFE (my intent
being to get rid of it entirely), but it takes time, and this particular
corner of the battlefield isn't one of the strategic positions I'm
focusing on. Once I manage to get close enough to the VFE's heart to
drive a stake through it, I expect this issue to become quite easy to
fix as well.

Post a reply to this message

> Am 02.02.2017 um 03:26 schrieb Cousin Ricky:

> >
> > Guess what?  The one thing those two scenes had in common that all my
> > other tests did not was Greek text.
> >
> > This appears to be the same charset utf8 bug that ThH uncovered.  And
> > since LanuHum probably uses a lot of Cyrillic text, I'm guessing that
> > his is the same problem.
>
> [Clipka wrote...]
> Sounds totally plausible.
>
> If that is the case, then the next beta should fix it.

> [Clipka wrote...]
>> It would be interesting to know whether the latest tagged alpha
>> (v3.7.1-alpha.8927145) works better than the beta. Since we've made only
>> very few changes to the source code since then, that would give us a
>> clear hint where to search for things that might be going wrong.
>
> [Cousin Ricky wrote...]
> Alpha.8927145 crashes.  Alpha.8917471 runs splendidly.

> [Clipka wrote...]
> That fits well with the theory that it's the "utf8" bug you're experiencing.

Sorry to be late with this, but...

I can confirm this  charset utf8  crash bug, in two versions of POV-Ray (on
Windows 7 64-bit):

Alpha.8927145
and
3.7.1-beta 2

Does NOT crash when using the 'original' v3.7.0 version.

(I used the appropriate pvengine64 versions for each.)

The scene that I did the tests with is mentioned here (I simply commented-out
the scene's 'charset utf8' entry in global_settings)...

http://news.povray.org/povray.beta-test.binaries/thread/%3Cweb.5897b07b6a1a359883fb31c0%40news.povray.org%3E/

and

http://news.povray.org/povray.text.scene-files/thread/%3Cweb.4a7c8c12a399eaf3df22a6900%40news.povray.org%3E/?ttop=41306
4&toff=50

Post a reply to this message

On 2017-02-05 07:47 PM (-4), Kenneth wrote:
> Sorry to be late with this, but...
>
> I can confirm this  charset utf8  crash bug, in two versions of POV-Ray (on
> Windows 7 64-bit):
>
> Alpha.8927145
> and
> 3.7.1-beta 2
>
> Does NOT crash when using the 'original' v3.7.0 version.
>
> (I used the appropriate pvengine64 versions for each.)
>
> The scene that I did the tests with is mentioned here (I simply commented-out
> the scene's 'charset utf8' entry in global_settings)...
>
>
http://news.povray.org/povray.beta-test.binaries/thread/%3Cweb.5897b07b6a1a359883fb31c0%40news.povray.org%3E/
>
> and
>
>
http://news.povray.org/povray.text.scene-files/thread/%3Cweb.4a7c8c12a399eaf3df22a6900%40news.povray.org%3E/?ttop=41306
> 4&toff=50

The latest master branch does not crash.  I look forward to beta 3.

Post a reply to this message