[Uml-devel] some answers....

[Luis De la Parra]

Hi all, some answers to some of you...

> Firstly, thanks for your code.  I'll get Luis who is looking after code g=
> en.=20
> to look at your code.

I looked a it last night and added it to my copy (with some modifications of
my own).
Thanks a lot. I now suggest to wait a day or two untill Paul can update CVS
and is stable, and then it should be easy to commit this new code

> > I was able to fix the problem with:
> > 
> >   if( &(m_List) != &(rhs.m_List) )
> 
> That fixes the occurance of the compile error, but is not right. I'd
> suggest 
> to use
> 
>   if( !(m_List == rhs.m_List) )

That is true, what I did might not be right, but I said it loud enough here.
I think several times, that I did that ugly fix only to get it to build, but
that I did not know if the author of that code wanted make sure the two are
actually the same object (by comparing only the addresses) or if he wanted to
compare the list contents. 
no one answered so I could only hope the one responsible for it saw it and
thought it was ok.

> When the new version 1.0.4 is released a little app will be released with=
>  the=20
> program that will convert the older file format.  I may even just impleme=
> nt=20
> in the program itself.  I am yet to write this.  Someone else may even wr=

I would suggest a separate utility... this tool is only going to be used
once for converting all files to the new format, so I dont think we should put
it into the program itself... it would then stay there forever.

> 
> KDevelop no longer make a uml binary but something like uml-i386-linux or=
> =20
> something like that.  This means program won't load from menu atm.

I reported that problem a while ago.. there's nothing wrong with the
program, but you (or KDevelop) are adding --target and --host to configure. just
take it away.. when you "configure" with KDevelop it shows this pop up for the
parameters to configure, well, make sure you only have the "prefix" and
"debug" options, you can take the rest out and then you will have a nice binary
called "uml"

> The CVS versioning in the download is not correct but it is the most up t=
> o=20
> date.  I will fix the versions when cvs is fixed and let you know.

so the download file is now the most up-to date? (including
uml-luis2.tar.bz2??) if yes, please confirm so that I can go get it.


> sourceforge. I noticed, that Paul changed the list comparison in 
> 
>   bool UMLOperation::operator==( UMLOperation & rhs )
> 
> to
> 
>   if( &m_List != &(rhs.m_List) )

dont blame him... that was me, but I warned that I did not know if this was
ok and that I only did it to make my copy of uml to build.. see the response
above.


> As one can see, the same two lists might have the same contents but
> different 
> addresses (as they are stored with different objects).
> 
> I would assume, that 3) is the correct result not 4).
> 
> Or am I missing something here? Please tell me if I am completely wrong, 
> because then I have to consider to go back to the books on C++ again.

no.. you are 100% right. I didnt write that code and just didnt have time to
look at the it deep enough to see if we were trying to just compare the
addresses, that means, make sure we are talking about the same object, or if we
wanted to compare the contents of the lists. I just wanted it to build.
there's nothing wrong with going back to the books once in a while though..
ther's always something new to learn!  =))


regards!

luis.

-- 
GMX - Die Kommunikationsplattform im Internet.
http://www.gmx.net