Re: [Tcllib-devel] ping gsoc graph work ...

[Andreas K. <and...@ac...>]

Michał Antoniewski wrote:
> Regarding mails before:
>  
> Crush and crash....yeah my mistake:) I went with wrong letter...:) thx:)

No problem :) Should you ever read fanfiction you will find that things can be 
very much worse when it comes to mangling the english language ... Even 
mistakes like canon/cannon role/roll etc. are quite harmless compared to some 
atrocities seen.


>  >>Now that I am focused on this I do however see one possible problem 
> with the naming. Consider two arcs X and Y, the first >>between the 
> nodes 'a' and 'bc' and the second between 'ab' and 'c.
>  >>       X => $u$v = 'a' + 'bc' = 'abc'
>  >>       Y => $u$v = 'ab' + 'c' = 'abc'
>  
> Oh, I accidentaly ommited that case... Right, the same problem was 
> considering Bellman's Ford, but this time I did one bad assumption and 
> lost it. Anyways, just changing keys to lists this time didn't do the 
> trick, because there was a problem during execution of isEulerian? 
> method.

Uh. That is then a bug in isEulerian?. We should be able to handle all types of 
node names.

 > Fleury method ( used by Eulerian ) returns error, when it's
> given such graphs.

Definitely sounding like a bug in the existing code.

 > I will have to take a deeper look into it as firstly
> I wanted to end the new TSP implementation. Probably, it's not a big 
> problem, will see....

Please file a bug report for isEulerian?/Fleury about this (*) _if_ the search 
for the bug in Fleury takes to much of your time.

(Ad *) At https://sourceforge.net/tracker/?group_id=12883&atid=112883

>  >>This command makes the command ::struct::graph available as ::graph
>  >>The test environment masked the bug.
>  >>As reminder for myself to get rid of this
>  >>https://sourceforge.net/tracker/?func=detail&aid=2811747&group_id=12883&atid=112883 
> <https://sourceforge.net/tracker/?func=detail&aid=2811747&group_id=12883&atid=112883>
>  
> Oh, right... so now it's obvious why tests went good everytime.... I'm 
> glad that by accident it helped to find a bug.

I'm glad as well.

>  >>Now I am reminded of RosettaCode, a nice site for exmaple tasks 
> implemented by lots of languages for comparison.
>  >>Tcl's page is
>  >>      http://www.rosettacode.org/wiki/Category:Tcl
>  >>And
>  >>      http://www.rosettacode.org/wiki/Tasks_not_implemented_in_Tcl
>  
> Interesting! Actually, now after spending some time coding in Tcl, I'm a 
> bit surprised, how it happened that Tcl was almost fully omitted during 
> classes at my University.... We was using Tcl scripts for example, when 
> doing some simulations on computer networks, but when it came to 
> subjects about programming in script languages, there was nothing about 
> Tcl...weird.

Tcl is sometimes called the best-kept secret among the scripting languages.
Companies don't want you to know the secret of their success, you might be came 
as effective as them in writing solutions in this easily used language. This is 
a bit tongue-in-cheek, but also has a kernel of truth.

Hopefully we, the community, manage to change that.

Andreas