Re: XMI

[Alan Horkan <horkana tcd ie>]

On Sun, 15 Sep 2002, James Michael DuPont wrote:

> > Compiling argouml with gcj would be an "interesting" approach.
> Just the core parts. There are a few nice Java libraries with few
> system dependancies. The Novosoft XMI engine might be interesting to
> link in.
> > You may have noticed i have a thing for code reuse and portability
> > and
> > cooperation, if you plan to do this in big way maybe you could get
> > some of
> > the Kivio developers to cooperate (random suggestion of the day).
>
> Or the Argo developers. One reason why I like dia over argo is that it
> is fast. Argo is a pig (But a sweet one, mmmmh : ham sandwiches come to
> mind)

you suggest using GCJ (Gnu Compiler for Java) have you used it on Argo
before and is it still slow even when compiled to a native linux binary
using GCJ?
is that other commercial (?) program based on Argo also slow?
being able to exchange files with Argo would be great.

> The kvivio is an insult to free software, never have I felt so insulted
> by a piece of software, before. Never have I downloaded such a
> shameless plug to make money.

Free software, what a misnomer!  (dont flame me please).

I assume you mean Kivio.
In the unlikely event there are any KDE developers on this list they might
get the mistaken impression that we all have a major problem with Kivio.

The General Public License is not incompatible with making profit.

> If I ever find the time, I am going to write lots of hatemail about
> this crippled software.

Kivio MP (the qt only version) has a very odd license, and i wonder how
different it is from the Kivio that is released under the GPL as part of
KOffice, i appluad the fact that they release source, and i appluad the
fact that they release at least some of it under an OSI approved license.
Kivio MP includes quite a few stencils.
Dont make the mistake of judging a project based solely on what the debian
packager decided to include, as i am sure you are well aware how quickly
things change in software.  I still hear linux users making complaints
about RedHat based on the really old version they used, or gurus
recommending mutt when evelotion or other newer 'easier to use' options
(subjective, i know but anway) are available.

As a commercial software project selling stencils does not seem like an
ideal source of revenue, and i hope after reading the rest of this you
will understand why exactly i think so.  A smarter revenue stream in my
opinion would be to charge for Import and especially export to proprietary
formats such as PDF, Rational Rose, or much harder Visio.  (writing
something like libvisio would be a hell of a challenge but if correctly
managed could actually be quite profitable).

Templates and stencils should be an area where there is a very low barrier
to entry (being able to hand edit XML help but not essential)  and we can
really get users involved and wipe the floor with proprietary software by
having a vast selection of templates.  One of my goals for Dia is to
encourage users to give us templates, and as a free software open source
project we should in theory be able to have a vast collection of
predesigned shapes.  The best user submitted templates could be cleaned up
and grouped with Dia.  There are things that i would do before making any
public appeal for ordinary user to start authoring templates.

1) there would need to be clear, easy and up to date instuctions on shape
building and design.  Convincing other apps to be able to export Dia
shapes would help shape design.
2) easy way to submit shapes, including a well organised plan how Dia
would manage to put new shapes if we had 100s of shapes in many differnt
categories (ambitious maybe)
3) easy way to add shapes, perhaps a zip file that users could simply
'Add' and have dia put the sheets and shapes in the appropriate directory.
(would need to be very careful to avoid the scary prospect of reinventing
package management).

> I installed the debian package, tried to create an UML diagram, and was
> told to Buy stencils from them, they did not even provide good samples.
> What a waste of time!
>
> Sketch on the other hand is nice, but also a bit slower than Dia.
> I would not mind working with them, but the Kompany will get no support
> from me.

> > Why do i keep doing this?  These emails and these links get archived
> > on
> > the web, and it affects a certain very popular search engine.  A few
> > months ago Dia was not the first listed item when you searched using
> > the
> > word Dia but it is now.  Link early, link often.
>
> Why not put the dia link in the mailman, that way it shows up on every
> single mail?

I dont control mailman.  but yeah its a good idea.

i would like if mailman could strip these kinds of advertising blocks tho'
i guess i should learn a bit of perl and write some pattern matching
scripts.  (its not like anyone signs webmail so there is no big deal about
damaging the integrity of the message).

> Do you Yahoo!?
> Yahoo! News - Today's headlines
> http://news.yahoo.com

another random brainfart i had this week was that when replying to a
mailing list digest i wish i could "Undigest" or "Regurgitate" (like that
word) the message and reply to a specific MIME part and that way have the
correct subject line and headers and my response would thread nicely to
the original message.  I should probably keep a software "wishlist" maybe
some one will be daft enough to actually develop one of my random ideas,
in the meantime i should probably file a bug http://buzillla.mozilla.org

So much for a short response, I am so easily distracted, I had better go
do some work ...

Sincerley

Alan Horkan
Dublin, Ireland.
http://matrix.netsoc.tcd.ie/~horkana
http://advogato.org/person/AlanHorkan/