Board index » delphi » Re: Delphi Bugs

Re: Delphi Bugs


2004-07-27 12:21:34 PM
delphi212
"Jeff Overcash (TeamB)" <XXXX@XXXXX.COM>wrote
Quote
Because Borland cares about doing little things
like regression testing which the freepascal people
obviously don't (or else have such a small set of
regression tests as to make them worthless).
Ouch! That statement makes me think of the amount
of testing required to catch the problem of function
Math.SimpleRoundTo()'s error with practically all
negative inputs. I suppose that if they used two tests
one with positive and one with negative input the but
would have been caught. But I suppose that that was
a rare slip up.
I guess that Borland has or could have a regression
testing operation that would accept a new set of builds
each day and automatically run all of the tests that have
ever been submitted by customers and internal
programmers that could apply to the builds.
The bugs that bother me are mostly code that has apparently
never worked correctly, like the divide function in FMTBCD
(QC#6013,#6063), the SimpleRoundTo (QC#8143), and the
use of Int64 control variables in case statements (QC#7784).
Regards, JohnH
 
 

Re: Delphi Bugs

More realistically, given what we saw with the D7SP1,
it's because Borland does NOT have a single regression
test and are afraid to break everything.
Z
 

Re: Delphi Bugs

Quote
If you don't want quality control then I guess the free pascal
method is for you, for most people they want some sort of QA
done on patches.
If you look at some Bugs in Delphi 7 - not resolved till now - there is the
question if Borland does something like testing at all...
--
cruxyATcruxyDOTnet
XanaNews 1.16.3.1
 

Re: Delphi Bugs

Quote
They fixed it, which is hardly "nothing". ;-)
It took over one year of "regression testing". In that case the Free
Pascal Method is "better". Better for the work to be done...
But that does not mean that I want to use FreePascal right now - the
project is not in a state to use it in production environment - IMO.
--
cruxyATcruxyDOTnet
XanaNews 1.16.3.1
 

Re: Delphi Bugs

Jeff Overcash (TeamB) writes:
Quote
"Max" <XXXX@XXXXX.COM>writes:

>Regression testing is a reason not to do it ????


So how many things did the "fix" break? If you use FreePascal
I guess you don't have any clue. Borland doesn't jsut throw
things out at their customers, they make sure it goes through
a complete testing process to minimize the possibility of
breaking other code with a fix. This takes time, definitely
more than 24 hours in the case of Delphi.
Laughable if you have ever worked with ECO and/or C#Builder. Borland has
obviously no problems at throwing things to their customers untested.
Meanwhile I know why Borland products are that high priced. You get much
more bugs per dollar than you would ever get from their competition.
By the way, I "like" Borlands cynicism on their splash screen:
Borland - Excellence Endures
 

Re: Delphi Bugs

Max <XXXX@XXXXX.COM>writes <410541dc$XXXX@XXXXX.COM>
Quote
Yesterday I have reported a freepascal bug. Today: its fixed! Bugs reported

qc are open since over a year! What is the reason that Borland does not fix
them?
A few years ago Danny Thorpe reported a very serious Linux bug. It still isn't
fixed. What is the reason open source projects do not fix bugs?
Danny's bug was a lot more serious than your's, so you lose. Goodbye.
--
***Free Your Mind***
Posted with JSNewsreader-BETA 0.9.3.220
 

Re: Delphi Bugs

"Nick Hodges [TeamB]" <XXXX@XXXXX.COM>writes
<41057003$XXXX@XXXXX.COM>
Quote
In other words, what assurance to you have that the "fix" doens't break
existing code in insidious ways?
<sarc>
But it is open source!! It must be perfect! After all, "the whole world" has
access to the code to fix it!
</sarc>
--
***Free Your Mind***
Posted with JSNewsreader-BETA 0.9.3.220
 

Re: Delphi Bugs

Cruxy <XXXX@XXXXX.COM>writes <XXXX@XXXXX.COM>
Quote
It took over one year of "regression testing". In that case the Free
Pascal Method is "better". Better for the work to be done...

But that does not mean that I want to use FreePascal right now - the
project is not in a state to use it in production environment - IMO.
That's because the FreePascal Method doesn't work. If it worked, you'd be
willing to use it in a production environment by now.
--
***Free Your Mind***
Posted with JSNewsreader-BETA 0.9.3.220
 

Re: Delphi Bugs

Bob <XXXX@XXXXX.COM>writes
<4105b0db$XXXX@XXXXX.COM>
Quote

I can sympathise -- I was nailed last year by a language bug (TRY-FINALLY in
functions can wipe out the return value) that was basic to the very
definition of the language: I'd consider a language bug to be of the
highest priority. Notwithstanding this, Borland has done nothing with it
so far (not even a caveat in the ReadMe documents).
Your sympathy is based on your own lack of information. That bug was fixed a
while ago.
--
***Free Your Mind***
Posted with JSNewsreader-BETA 0.9.3.220
 

Re: Delphi Bugs

Cruxy writes:
Quote
the
project is not in a state to use it in production environment - IMO.
Hmm.
--
Nick Hodges -- TeamB
Lemanix Corporation -- www.lemanix.com
Read my Blog -- www.lemanix.com/nick
 

Re: Delphi Bugs

"Nick Hodges [TeamB]" <XXXX@XXXXX.COM>wrote
Quote
>is not in a state to use ... in production environment
Hmm.
Reminds me of FMTBCD. Thank goodness, it is not used
very much, and the few programmers that have used it can
avoid the routines that do not work (e.g. divide).
 

Re: Delphi Bugs

"Captain Jake" <johnjac76[nospam]@comcast.net>writes
[..]
Quote
<sarc>
But it is open source!! It must be perfect! After all, "the whole world"
has
access to the code to fix it!
</sarc>
LOL
 

Re: Delphi Bugs

On Tue, 27 Jul 2004 09:12:44 -0500, "Captain Jake"
<johnjac76[nospam]@comcast.net>writes:
Quote
>But that does not mean that I want to use FreePascal right now - the
>project is not in a state to use it in production environment - IMO.

That's because the FreePascal Method doesn't work. If it worked, you'd be
willing to use it in a production environment by now.

I've been using FPC code in production for about 5 years. When I
started, the FPC principals themselves helped me get up and running.
The code has been running on several servers for years with zero
problems.
I don't think FPC is the best tool for Win32 development, primarily
because I am accustomed to the polish of Delphi.
For linux/32 development I am more comfortable with FPC than Kylix. I
have more confidence in FPC being supported over time.
For 64bit development I am more comfortable with FPC than Delphi/Kylix
for obvious reasons.
Quote
That's because the FreePascal Method doesn't work.
This is no less absurd than the original post. It obviously does work
in some situations. Zealots either for or against OSS are equally
laughable.
 

Re: Delphi Bugs

On Tue, 27 Jul 2004 09:09:31 -0500, Captain Jake writes:
Quote
"Nick Hodges [TeamB]" <XXXX@XXXXX.COM>writes
<41057003$XXXX@XXXXX.COM>
>In other words, what assurance to you have that the "fix" doens't break
>existing code in insidious ways?

<sarc>
But it is open source!! It must be perfect! After all, "the whole world" has
access to the code to fix it!
</sarc>
Jake just took another shot at OSS -- nothing new there.
 

Re: Delphi Bugs

On Tue, 27 Jul 2004 09:06:41 -0500, Captain Jake writes:
Quote
A few years ago Danny Thorpe reported a very serious Linux bug. It still isn't
fixed.
Evidence?