Re: The early years
- From: "Ralf Mimoun" <nospam@xxxxxxxxx>
- Date: Mon, 11 Jul 2005 22:41:14 +0200
James K Smith wrote:
[historical Kahn quote]
All the quality discussion / flame war, the list of what Delphi should
become, the last DevTracks I attended, the assignments I got in the last
yeats and my experience made me think. Happens not that often. Oh, and
everyhing here is 100% personal opinion. I don't want to say that it's the
way Borland should go to be successful. It's just my limited view.
I am working with D5 Enterprise. Yes, it's stone old, but it is reliable. I
have some tools to help me with the daily work, and I have a heap of 3rd
party components. Years ago, I bought the upgrade to D6 after a Borland
product show, but I tried it only for 10 minutes - simply instable. That
money went down the drain, ROI was zero. The comment of Borland customer
care was amusing: I could give if back it I did not open it. But I had to to
see if it's useable. They were sorry. So was I.
Why don't I upgrade to D2005? Or better, why should I? I don't really need
ECO, I don't even use UML. It does not help my customers; I did not run in a
situation where another layer would have saved time. But what my customers
know: there is no such thing like a bug free application, and they know that
I will deliver a bug fix asap, normally in hours. My best tool here is
madExcept - and it's a shame that Delphi does not have something similar.
Same for FastMM4 debug features, it could be in Delphi since version 5. At
least.
The only original visible VCL "component" I use in newer projects is TForm.
The rest is DevExpress and Raize. Almost every single component shipped with
Delphi looks old and dusty (ok, except maybe TLabel), and one of the
wonderful glyFX glyphs one a TBitBtn does not fit in my eye.
Then "the rest": reports are done with ReportBuilder for the complicated
stuff and ExpressPrinting for lists (with grouping, filters, you know
EQGrid). Database? DBISAM if possible. Setups are done with Inno Setup. 3
tier? kbmMW. Then PAL for source code quality control. Wow, that's some
money, plus all the upgrades to new versions every year or so. Btw,
something like PAL and Pascal kinda LINT, integrated in the IDE with direct
feedback would make me much more productive. If I write questionable code
that could lead to a bug, why should I wait 6 months for the first bug
report if the IDE could see it while I write that line?
My customers order more or less the same kind of applications that they did
10 years ago. Financial stuff, db related. The differences: it has to be
more powerful (mostly user defined list of all kind), faster, with a better
and nicer UI, and more export formats. More features, more feedback, more
integrated in the daily workflow. But it's the same kind of stuff. 3 tier is
very rare, and once a year we have to write a small DCOM server. Same for
web related services. XML? They want it, but they don't need it full
flavoured. I am talking about large companies, and all they do is very
simple XML export. But what I need is access to different DB systems:
Oracle, MS SQL, DB2 etc.
It is great that Delphi 5 is so extensible, but don't get me wrong. I want
to invest in a new Delphi license, but it should help me, raise my
productivity. I don't need ECO (or better, I don't see a benefit for me
right now), but what I need is full featured refactoring, a folding editor,
much better compiler optimization, Win64, compiler support for newer CPUs
(SSE is not that new), build tool (like FinalBuilder, yes, I could buy that,
too). The IDE must (not should) be rock solid, not less. I don't want to
wait, don't want to restart. The tool should be made of stainless steel, or
better titanium. Oh, and fast. If I want to change a property or jump
between forms and code, I want to do something, get my work done. Don't
interrupt me with flickering forms, just let me do what I want. If I ever
deliver something to my customers that seems to be unuseable whan all
features are turned on, I would be out of business in months - or I should
fix that. No extra cost, I have to make it happen. Not my fault, my customer
demand that a feature does what it should.
Part of RAD is using a high abstraction layer, eg. OO, ECO, design patterns,
components. Another part is using a tool that lets you do what you want
immediately. Without a thought why something needs another second or if the
IDE kicks you out in the worst moment possible. So, I am willing to pay for
every product that make the daily work faster, that catches bugs as soon as
possible, that make my customer happy.
I know that I am only one customer, and there are other with different
needs. I am running a relative small software company, so I have relative
small requirements. As long as Borland does not deliver what I need, I don't
see a reason to upgrade. Don't get me wrong, I am sure that Borland knows
what they do. It simply seems that Borland today does not fit to me anymore.
Or just right now, I hope.
Wow, I did not intend to write that much. Maybe I should simply state:
Personally, I don't want a nano tech robot plant. Just give me a better
Leatherman tool.
Ralf
.
- Follow-Ups:
- Re: The early years
- From: Adam Roslon
- Re: The early years
- From: steve-o
- Re: The early years
- From: Dave White
- Re: The early years
- References:
- The early years
- From: James K Smith
- The early years
- Prev by Date: Re: Catastrophic Failure Trying out VCL.NET
- Next by Date: Re: VCL (Win32) in Longhorn ?
- Previous by thread: The early years
- Next by thread: Re: The early years
- Index(es):
Relevant Pages
|
