[okular] [Bug 383915] Okular crashes with a segfault on reload for some synctex files

[Flupp]

https://bugs.kde.org/show_bug.cgi?id=383915

--- Comment #19 from Flupp <bugs.kde.org at derflupp.e4ward.com> ---
Created attachment 109133
  --> https://bugs.kde.org/attachment.cgi?id=109133&action=edit
recursively search for pdf with synctex in current dir and open in okular

The commit in comment 18 seems to fix all my synctex-related crash problems. In
fact, I have opened every pdf accompanied by a synctex file I could find in my
home dir (cf. attached script) and no error occurred (246 files tested). For
some files I got several messages of the form

! SyncTeX Error : Ignored record ...>(line ${some_integer})


@comment 16:

> Ideas:
I have another idea for working around issues with synctex: Okular could just
call the synctex command line tool with the edit subcommand as soon as the user
shift-clicks. While this is still not a perfect solution, it has the following
advantages:

* Okular does not have to know anything about synctex – it does not even have
to check, if there is a synctex file, although this would be more user
friendly.

* Loading of the synctex file (and therefore execution of a potential exploit)
is delayed until the user really asks for synctex by shift-clicking.

* If the synctex cmdline tool crashes, it does not affect okular (but now the
synctex cndline tool might be exploited, of course).

* After the first shift-click, the user could be asked for confirmation, if he
trusts the synctex file.

* Okular would not depend on synctex. Okular would work just fine, if the
synctex cmdline tool is not installed.

Delaying to load the synctex file until the first shift-click is of course also
an option if synctex is used as library.


> - In some situations, the synctex upstream code shows parse errors.
> 
> Flupp: Could you report [this] first problem over at https://github.com/jlaurens/synctex/issues?
Are you sure, this is not an expected behavior for old synctex files?


@comment 17:

> No idea why sometimes it does not crash for Flupp, perhaps sheer luck with some pointer access.
I’d guess the same. The guess is supported by the random characters appearing
in the output.

-- 
You are receiving this mail because:
You are the assignee for the bug.