Re: [Tcllib-devel] Graph Manip

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

Michał Antoniewski wrote:
> Another update is available at assembla.
> What's new:
> - new test cases ( regarding your treatise about tests, I updated some 
> missing cases )
> - some of results ( for test cases ) moved into procedures - the bottom 
> of XOpsSetup file, I wanted to stick with organisation given by XSetup 
> file for "setup files".
> - updated description of algorithms and code comments - graphops.tcl.
> - corrected errors you noticed in last version.

Ok. Will review sometime today.

> First of all, thank you for your treatise concerning tests - it has 
> surely well described approach to tests for all algorithms 

Thanks, glad that it was of help to you.

> implementations. So, in each of next steps, I'm going to stick to 
> black/white box testing principles.

>  >>So, if you now believe that writing tests for an algorithm takes at 
> least as much time as coding the algorithm, and often more time, then 
> you are completely right. It does.
>  
> Yeah, thats true:) I would even say that it takes a lot more time than 
> implementing, especially concerning easier implementations.

> This year, D. Richard Hipp had a lecture at my university and when he 
> said about proportions of test code and other code in SQLite 
> implementation... let's say I was a little bit amazed by this huge 
> disproportion :)

Heh.

sqlite is an extremely well tested piece of software
(http://sqlite.org/testing.html).

And Richard a very, very smart person. Miles ahead of myself.

>  >>Do you have web references about the algorithm, at wikipedia or other 
> website ?
> Yes, I found algorithm descriptions at wikipedia and placed into code. 
> As I'm mostly using Polish sites for theoretical info about graphs ( 
> I've got some favourites where I can find quickly everything I need :) ) 
> it's not a good idea to put those references. So, wiki websites in 
> English are used.

Thanks, and agreed that most of the audience will be lost on a Polish site.

> Identation:
>  >>May I ask what editor you use ?
> When I'm working on my notebook, I'm using Notepad++, so maybe that's 
> the cause.

Well, I do not know Notepad++ so I can't say.

 > Otherwise, I use Vim.

I seem to remember that VIM has a Tcl mode, or some such ...
Googled vim + tcl, found
	http://www.vim.org/scripts/script.php?script_id=1603

That is syntax highlighting however, not indenting ...

Aha, http://www.vim.org/scripts/script.php?script_id=178
says something about indenting.

Note: While I found these files I have not used them in any way, shape or form. 
  I.e. I have no idea about the quality of these.

	Maybe some other other readers of tcllib-devel and using vim
	will wish to jump in on this ?

> About returning values:
>    -  I changed it and now it's set to return error when there is found 
> any negative cycle.
>    -  There is no risk now, that deleting node won't happen.
>    -  Generally, graph isn't changed at all by procedure - it relates to 
> nodes as well as weights of edges.

Ok.

> About -cutdisplay option, and $param variable:
>    -  yes, you're right - those names are bad.....I just hadn't good 
> idea how to name them

Right, that can happen. And going forward is better than stalling, naming is 
something which is of a lower priority compared to most other problems.

>    -  However, I came up with option -filter ( it is used by other 
> struct::graph procedures, for limiting the set of returned values e.g. 
> nodes - this name will assure some logical conformity ), and $param is 
> now $displaymode.

Hm. No comments regarding this yet, will have a look at the code first.

>  >>I found only a quibble and a typo ...
>  >>'...UNCOHERENT...' in XOpsSetup ...
>  >>How about '...PARTIALLYCONNECTED...' instead.
>  >>Thats the quibble.
>  
> I named it uncoherent, because coherence is in Polish term for 
> [strongly] connected graphs. In most cases translating this way works, 
> but this time it turned out that not always... :)

Very true.

A similar translation problem a german coder could have is 'aktualisieren'. 
Might be tempted to translate as 'actualize'. Wrong. Correct translation is 
'update'. 'actualize' in german is 'verwirklichen'. Completely different meaning.

> It is set now to PARTIALLYCONNECTED.
>  
> Since tomorrow I will start implementing next algorithms and also some 
> updating of existing solutions to come closer to ending first's week phase.

Ok, looking forward to them.

Andreas.