Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.misc > #6876 > unrolled thread

a look at some hard-to-spot bugs in Libreoffice code

Back to article view | Back to comp.misc

Contents

  a look at some hard-to-spot bugs in Libreoffice code RS Wood <rsw@therandymon.com> - 2015-03-02 14:09 +0000
    Re: a look at some hard-to-spot bugs in Libreoffice code pnwcha@black.box - 2015-03-02 18:20 +0000
      Re: a look at some hard-to-spot bugs in Libreoffice code RS Wood <rsw@therandymon.com> - 2015-03-03 13:44 +0000
      Re: a look at some hard-to-spot bugs in Libreoffice code Huge <Huge@nowhere.much.invalid> - 2015-03-03 22:14 +0000

#6876 — a look at some hard-to-spot bugs in Libreoffice code

http://www.viva64.com/en/b/0308/#ID0ECRAI

It's essentially publicity for the tool used (see the conclusion at the
end), but it's interesting, nonetheless.

LibreOffice is free and open-source. It includes code written in Java,
Python, and C++. We analyzed the part written in C++ (and a small part
in C, C++/CLI). Version: 4.5.0.0.alpha0+ (Git revision: 368367).

Analysis was done with the static code analyzer PVS-Studio.

Let's have a look at the errors found in this project and see what can
be done about them. I'd like to notice right away that some of what I
think to be bugs may actually be no bugs at all. As I'm not familiar
with the code, I could have easily mixed up a real defect and correct
code. However, since such code confused both me and the analyzer, it
certainly means something isn't right there. This code smells and
should be refactored to reduce the probability of its being
misunderstood in the course of project development and maintenance.

* Typos
* Code probably written on purpose but still looking suspicious
* Copy/Paste
* Brave use of the realloc() function
* Logical errors
* Skeletons in the closet
* Safety errors
* Misc

Conclusion

Although the number of errors, defects, and slip-ups discussed in this
article is, as usual, great, the LibreOffice project's code is still
very high-quality. And it does bear the evidence of being regularly
checked by Coverity, which indicates the authors' serious approach to
the development. The number of bugs found by PVS-Studio is pretty small
for such a large project like LibreOffice.

What did I mean to say by this article? Well, nothing special, really.
It's a bit of advertising for our tool, and that's all. Use the
PVS-Studio static analyzer regularly to find and fix piles of errors at
the earliest development stages.

-- 
RS Wood <rsw@therandymon.com
Posting to comp.misc, sci.misc, and misc.news.internet.discuss.

[toc] | [next] | [standalone]

#6880

>Let's have a look at the errors found in this project and see what can
>be done about them...

Heh, they don't do much; there's a really old bug dating back to the Star
Office days that affects files saved in the .doc/.docx format - basically
you can't save a document with a starting page # greater than 255 w/o the
page # reverting to "1".  Maybe it's just a tough bug to fix but it seems
like something that should be gotten to since most of the publishing
industry wants book files submitted as an MS Word document.  If you're
writing a book you probably have the chapters in separate files;
maintaining the page numbering is kind of important.  Both the LibreOffice
and OpenOffice projects have this bug filed but have yet to fix it; guess
they are hoping MS Word will just go away.

[toc] | [prev] | [next] | [standalone]

#6889

On Mon, 2 Mar 2015 18:20:32 +0000 (UTC)
pnwcha@black.box wrote:

 Maybe it's just a tough bug to fix but it seems
> like something that should be gotten to since most of the publishing
> industry wants book files submitted as an MS Word document.  If you're
> writing a book you probably have the chapters in separate files;
> maintaining the page numbering is kind of important.  Both the LibreOffice
> and OpenOffice projects have this bug filed but have yet to fix it; guess
> they are hoping MS Word will just go away.

I think that's the general annoyance in opensource projects in general
- unless someone on the dev team really cares about the bug, it can
last for ages.

[toc] | [prev] | [next] | [standalone]

#6892

On 2015-03-02, pnwcha@black.box <pnwcha@black.box> wrote:
>>Let's have a look at the errors found in this project and see what can
>>be done about them...
>
> Heh, they don't do much; there's a really old bug dating back to the Star
> Office days that affects files saved in the .doc/.docx format - basically
> you can't save a document with a starting page # greater than 255 w/o the
> page # reverting to "1".  Maybe it's just a tough bug to fix but it seems
> like something that should be gotten to since most of the publishing
> industry wants book files submitted as an MS Word document. 

Fuck knows why. Word sucks for that.

-- 
Today is Boomtime, the 62nd day of Chaos in the YOLD 3181
I don't have an attitude problem.  If you have a problem with my attitude,
                    that's your problem.

[toc] | [prev] | [standalone]

Back to top | Article view | comp.misc

csiph-web