[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Renaming 'math' section 'science'




On Wed, 23 Sep 1998 treacy@debian.org wrote:

> > trd>You will need to get people's attention by starting a new thread called
> > trd>"rename math directory to science" or something like that.  Not enough
> > trd>people are reading this discussion.
> > 
> There is a real problem here that different people have completely different
> ideas on how things should be organized. For example, I consider most of
> the entries under math to be concerned with numerical analysis (fft in fftw;
> linear algebra in meschach, gsl (this may do more), saml; numerically solving
> diff. eqs. in felt; general numerical computations in octave and scilab).
> Others with a less numerical background may consider this all just to be
> 'science'. Others who don't even know what this stuff is may think it is
> all 'math'.
> 
> Here is I feel 'math' should be broken up:
> num_anal:
>    fftw
>    meschach
>    gsl
>    saml
>    felt
>    octave
>    scilab
>    latex
>    cln
>    python-numeric
>    pretty much anything released by SAL
> 
> math:
>    apcalc
>    all calculators (this is where most people would expect to find them)
>    bc
>    calc
>    computer algebra systems (no free ones yet that I know of)
>    statistical packages (or num_anal depending on the background of
>               the statistician. I'll let them decide)
> 
> graphics:
>    all plotting programs
> 
> ???:
>    hmmmer
>    spreadsheets
> 
> Note that I don't know what everything is so haven't categorized it all.
> 
> 
> Actually, in thinking about this, the whole system of categorizing packages
> needs to be redone in Debian. How packages are organized for selection by
> users should be much more flexible than our current system allows. What I
> am thinking is to expand the 'Section' variable in the Packages files to
> include subsections. This won't affect how the archive is organized, but
> in the long run will make package selection much easier.
> 
> This would allow us to have categories like the following:
> misc/dbserver        all database servers
> misc/dbclient        all database clients
I think database stuff should have it's own top level section.

> math/num_anal        numerical analysis software
> math/calculator      all calculators
> 
> New categories could only be added if the policy group allows it.
> This method of organizing would allow us to create new categories and
> get rid of old ones without having to reorganize the archive.
> For example, if it was decided that num_anal was too broad, it could
> be broken down into math/lin_alg and math/num_integration.
> Thus, the existing bad section names wouldn't matter as the user won't see
> them.
> 
> This would enable the user to go into an improved version of dselect or apt
> and instead of being hit by the entire section 'web' could see the sections
> 'web browsers', 'web servers', 'html editor', 'web log analyzer' and 'CGI'
> and have many fewer packages to select from.
> 
> There are already some minor flaws I've noticed in this which are easily
> remedied. If people like the idea I could expand upon it and bring it to
> -policy. The big stumbling block to this would be the necessary changes to
> the archive management programs, dselect and apt.

I like this idea, but what about things like perl interfaces to databases.
Should they go in databases or interpreters (interpreters/perl)? or should
they go in both.  And then things get really ugly.
							- Tom


Reply to: