Review Request: Fixing bug 204068, adding ERROR-Token to parser

[Albert Astals Cid]

> On Dec. 11, 2011, 10:58 p.m., Albert Astals Cid wrote:
> > I am not sure (3) makes sense. Also would it be possible for you to send two separate patches, one with just the spacing and the other with the feature? It would make it really easier to review, since now you have to look twice to see if it's just whitespace changes or also there some code change in a given line.
> > 
> > Sorry for taking so much to answer but kmplot development is basically stalled and i just found your review request while doing a rourinary "janitor" review request round.
> 
> Albert Astals Cid wrote:
>     Hello, you there?
> 
> Eike Krumbacher wrote:
>     Hi! Sorry for not answering such a long time. Well, on kde-edu mailing list, I read about dropping kmplot and integrate all the features in KAlgebra. So further investigations into kmplot does not seem to be welcome that much. I think it is better to drop this request. What do you think?
>     
>     Eike
>

Yes, there are plans of dropping kmplot and integrating it into KAlgegra (this will happen not before 4.9), but what does this have to do with not fixing a crash bug that is in 4.8?


- Albert


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
http://git.reviewboard.kde.org/r/103182/#review8882
-----------------------------------------------------------


On Nov. 18, 2011, 11:26 p.m., Eike Krumbacher wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> http://git.reviewboard.kde.org/r/103182/
> -----------------------------------------------------------
> 
> (Updated Nov. 18, 2011, 11:26 p.m.)
> 
> 
> Review request for KDE Edu.
> 
> 
> Description
> -------
> 
> Hello Folks!
> 
> If you have a set of functions like
> f(x) = g x²
> g(x) = 2 x
> 
> * kmplot crashes without notice, because the parser reads "g" as a function and reports "Wrong number of arguments" but crashes.
> * If "g" is meant to be a predefined constant the parser crashes because g is read as a function with wrong argument counts. 
> 
> So there are several cases:
> 1) g(x) and f(x) exists in a file, reading the file should not crash
> 2) if g is a constant, whether in the file or "global", f(x) should read this constant first
> 3) a function set like 
> g(x) = 10 * g
> f(x) = g(x) + g
> should be read and displayed.
> 
> With the given diff, we can archive (1) and (2) but not (3), which now does not crash anymore but  gives strange results. In the surrounding of the changes, I added some whitespaces to reflect the style guide of the given code. The main part is an introduction of the token "ERROR", adding this token to the stack in case of an error. 
> 
> Have fun
> 
> Eike
> 
> 
> This addresses bug 204068.
>     http://bugs.kde.org/show_bug.cgi?id=204068
> 
> 
> Diffs
> -----
> 
>   kmplot/parser.h c3fb92e 
>   kmplot/parser.cpp 23569aa 
> 
> Diff: http://git.reviewboard.kde.org/r/103182/diff/diff
> 
> 
> Testing
> -------
> 
> (1), (2) and (3) was tested. The original file from the bug report associated with this review was tested. No crashes in these special cases. 
> 
> 
> Thanks,
> 
> Eike Krumbacher
> 
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/kde-edu/attachments/20111230/8ce68ac4/attachment.html>