On Jul 21, 3:12 pm, llothar <llot...@[EMAIL PROTECTED]
> wrote:
> I see improvments with ISE/Eiffel Software but i don't see
> improvements with SmartEiffel. There the situation is getting worse on
> every release. Did you read the comments about the latest release?
> Something like a new 10x slowdown on an already impossible to use
> system?
>
>
I will have to look at the post because I did not see it. It would
be interesting to understand the cause. That is, I can understand
the compilation speed decreasing because more checks are being
made (according to a poster), but I can not see the runtime speed
decreasing.
> Hmmm, difficult. From financial aspects it was a complete insane
> failure
> to go with platform independence when 95% of the world is using
> Windows.
> There is just no money in this game that justifies Eiffels platform
> independence.
I'm not so sure. Linux is improving a great deal. Now, this might
not help your specific market, but it could.
>
> Technically, yes it helped me. Now there is a surprise for all the
> Eiffel fans:
> Do you know where the benefits come from? It is jsut one a small part
> of DBC, something
> that you can also do with assert() in C (i have large parts of C with
> a 1:3
> rate, i mean 1 assert each 3 statements). It is first the idea and
> mentality and programming style of eiffel that helped me and second
> the good stack trace feature.
> No invariants no postconditions. Because you simply can't use them in
> real world applications, it would slow down your application to much.
> Only very simple preconditions are acceptable (if i wasn't drunken i
> think i could explain you why all Eiffel vendors make it wrong - well
> they all do what ETL is telling them, so the only one you can blame is
> Bertrand Meyer for his ideas). But nevertheless after two years i can
> say that my eiffel code is almost bug (crash) free, it is just the C++
> code and the OS/external App layer that crashs from time to time. I'm
> not sure if i could say this if i had used any other language.
I understood most of this paragraph except for the last sentence.
Could
you not have learned eiffel and then coded your software in another
language using the concepts you learned from eiffel? Or are you
saying
that you could not have bug free software unless you use eiffel?
Answering
yes to the second sentence is a powerful advertisement for the eiffel
language.
> Because the reasons have to be very im****tant to move a project from
> language A->B.
> At the moment there is no such difference. Imperative languages are
> quite the same
> no matter whats the language.
>
Agreed that there has to be im****tant reasons. The point was just
that these clients choose the best language at the time. And later
they find that much of the excitement about the language was hype.
I'm sure most people remember when Java first came out. This was
the "write once, run everywhere" phase. I was working on a project
at the time that needed cross-platform development. 2-3 vendors who
sold
(US) $10,000 to $20,000 licenses per seat for cross platform
development
tools went out of business. Some specific companies include Visix and
to a lesser degree an expert system company who's name I cannot
remember
at the moment, but who went back to the rule-based development market.
Then there were renamed Blaze Software and have been sold a few times
now.
> I've haven't read Meyers OO book before i started using Eiffel, i read
> "Eiffel the Language 2nd Edition" which is completely different, it's
> the definition not the marketing. My fault was that in the year
> 1999/2000/2001 Smalleiffel seems to be great, making huge improvements
> every month and you couldn't see what insane future was comming. I
> whish i would have followed Uwe Sander, the second most serious
> SmallEiffel user on this planet, in 2002 when he realized that
> SmallEiffel is an impossible way for serious software deveplopers -
> but i wasn't so clever. I had to pay the price but luckily it wasn't a
> complete knockout, but it is still a serious problem.
I remember reading posts about the elj project and I actually
downloaded
the SmallEiffel windows distribution to try it out. I liked the fact
that
the package was focuse on Windows and it worked well (for me). I am
not an
apologist for the SmartEiffel team, but since that time they have
seemed to become more open in their development style with SVN access
to their code. The number of libraries has increased, etc. Still, it
is a research project with no paying users (that I know of). This
fact does not get them completely off the hook, but it does explain
why they cannot operate as a more typical open source project. Their
revenues come from research grants. But,
people have the freedom to fork the project if they wish.
> When i look at the available libraries i realize why sticking with the
> mainstream is a good thing.
Again, the large clients I see are always struggling. It seems that
they
spend a lot of time with Java evaluating different frameworks. Some
times
I think it would be simpler and more productive to either build your
own
frameworks (if the base libraries are there) or work to improve 1 or
2
frameworks instead of evaluating a 20 or 30.
Thanks for being patient with me and answering my questions. It does
help me address some of my open issues.
Lee
|
