Board index » cppbuilder » Re: Most annoying problems with BCB2006

Re: Most annoying problems with BCB2006


2006-06-25 04:19:44 AM
cppbuilder53
In article <xn0env9iyhimnz008-velthuis@www.teamb.com>,
"Rudy Velthuis [TeamB]" < XXXX@XXXXX.COM >wrote:
Quote
>Wow. That's a nice put-down.

Was it? It wasn't meant to be.
I believe that you didn't mean it as such, but the implication was:
Sure, they might lie to *you*, but I'm better/more important/better
connected than you are, and they wouldn't lie to me.
It might have been phrased better as: You might argue that they might
want to deceive the public at large, but it would be unwise to deceive
TeamB because it would undermine the value that they get from TeamB.
--
-David
Quis custodiet custodes ipsos?
 
 

Re:Re: Most annoying problems with BCB2006

At 22:19:44, 24.06.2006, David Dean wrote:
Quote
It might have been phrased better as: You might argue that they
might want to deceive the public at large, but it would be unwise to
deceive TeamB because it would undermine the value that they get from
TeamB.
Right. That is what I meant. Thanks.
--
Rudy Velthuis [TeamB] rvelthuis.de/
"Not everything that can be counted counts, and not everything that
counts can be counted." -- Albert Einstein
 

Re:Re: Most annoying problems with BCB2006

In article <xn0enva70ihg1x00b-velthuis@www.teamb.com>,
"Rudy Velthuis [TeamB]" < XXXX@XXXXX.COM >wrote:
Quote
Right. That is what I meant. Thanks.
You're quite welcome.
--
-David
Quis custodiet custodes ipsos?
 

{smallsort}

Re:Re: Most annoying problems with BCB2006

"David Dean" < XXXX@XXXXX.COM >wrote in message
Quote
In article <449d67e2$ XXXX@XXXXX.COM >,
"Dennis Jones" < XXXX@XXXXX.COM >wrote:

>There is *no way* Borland could
>possibly use BCB for anything significant without encountering the same
>problems and bugs the rest of us do on a daily basis.

Daily basis? If you run into them that often, then you should have
been able to come up with a reproducible test case. Can you give us an
example?
The "we" was meant to include all of us public Borland users. There is no
doubt in my mind that every day some user encounters a bug that is a
relatively common occurence, not only for him, but quite likely many others.
If Borland used their own tool as much as the typical Borland user (by that
I mean the typical developer who frequents these newsgroups), surely they
(Borland) would also encounter many, if not all, of the same bugs, and would
therefore be much more likely to correct them.
- Dennis
 

Re:Re: Most annoying problems with BCB2006

In article <449da6db$ XXXX@XXXXX.COM >,
"Dennis Jones" < XXXX@XXXXX.COM >wrote:
Quote
If Borland used their own tool as much as the typical Borland user (by that
I mean the typical developer who frequents these newsgroups), surely they
(Borland) would also encounter many, if not all, of the same bugs, and would
therefore be much more likely to correct them.
I don't know how much the Borland R&D folks use the C++ personality,
but sometimes no matter how many times you use an app, you miss some
bugs. I know it happens to me. I test, test, test, and still fail to
discover bugs until the app is put into production. Maybe they have bad
luck, maybe they need to use the C++ personality more, maybe their
habits just coincidentally avoid the bugs that some people run into.
QA is a tough job, New {*word*76} might help, more time in the C++
personality might help, but good test cases will definitely help.
--
-David
Quis custodiet custodes ipsos?
 

Re:Re: Most annoying problems with BCB2006

"David Dean" < XXXX@XXXXX.COM >wrote in message
Quote
In article <449d67e2$ XXXX@XXXXX.COM >,
"Dennis Jones" < XXXX@XXXXX.COM >wrote:

>There is *no way* Borland could
>possibly use BCB for anything significant without encountering the same
>problems and bugs the rest of us do on a daily basis.

Daily basis? If you run into them that often, then you should have
been able to come up with a reproducible test case. Can you give us an
example?
On a daily basis I get internal compiler errors. I already used to making random
changes in one header file after which they usually go away. Almost on a daily
basis I have IDE crashes on start or exit. On a daily basis, when I do
debugging, the watches refuse to show the variables because de{*word*81} decided
they were not in the scope or didn't exist. There are plenty of other smaller
annoying bugs that surface every day; everybody who works with the product on a
daily basis already knows that. It's obvious, had Borland built a single medium
size project with BDS they would have ten page long list of defects.
In my view, it's completely ridiculous asking customers purchased commercial
product to do alpha-testing for them. Firstly, they should have done it *before*
the release. Secondly, if they want my company to do QA for them, they will have
to pay; I'm not fiddling with BDS out of curiosity or because I have nothing
better to do.
J
 

Re:Re: Most annoying problems with BCB2006

At 23:55:49, 24.06.2006, David Dean wrote:
Quote
I don't know how much the Borland R&D folks use the C++ personality,
but sometimes no matter how many times you use an app, you miss some
bugs. I know it happens to me. I test, test, test, and still fail to
discover bugs until the app is put into production.
Especially since other people's setups may be quite different.
--
Rudy Velthuis [TeamB] rvelthuis.de/
"I love Mickey Mouse more than any woman I have ever known."
- Walt Disney (1901-1966)
 

Re:Re: Most annoying problems with BCB2006

In article <449dbf36$ XXXX@XXXXX.COM >,
"J Alexander" < XXXX@XXXXX.COM >wrote:
Quote
everybody who works with the product on a
daily basis already knows that.
I use it every weekday, and I don't encounter any problems like you
mention with any regularity. I do work on many projects of varying
sizes. I do encounter problems. I've submitted dozens of AIRs, but they
sure don't seem like issues that could be solved as easily as you imply.
--
-David
Quis custodiet custodes ipsos?
 

Re:Re: Most annoying problems with BCB2006

David Dean schrieb:
Quote
In article <449d67e2$ XXXX@XXXXX.COM >,
"Dennis Jones" < XXXX@XXXXX.COM >wrote:

>There is *no way* Borland could
>possibly use BCB for anything significant without encountering the same
>problems and bugs the rest of us do on a daily basis.

Daily basis? If you run into them that often, then you should have
been able to come up with a reproducible test case. Can you give us an
example?
IMHO test cases are not appropriate for errors that occur only in
combination with large projects. The best way to get information about
an error or instable state in an application are dump files. Windows
Vista supports them directly. Right click on an applications entry in
the task manager and select "Create dump files". Then Windows will write
the current state of the application to disk.
This dump file may be loaded into a de{*word*81}, e.g. WinDBG and the
current state of the application can be debugged, variables, threads etc.
The same applies to errors like access violations. Windows (Vista, XP,
Win2K....) will start an error reporting tool or Dr. Watson on such
errors and will send a dump file to Microsoft. AFAIK you can register
your application, so that the dump file will be sent not to Microsoft
but to your company directly.
I've integrated dump file support into our own applications. They are
written on any errors or on demand, e.g. if the application hangs or got
into an instable state. Our customers will then send the dump file to us
for evaluation and we can evaluate the problem directly in the de{*word*81}.
No need to reproduce the problem.
The problem with BCB6 is that it doesn't support dump files directly. I
cannot load them into BCB's de{*word*81}, I have to use Microsoft de{*word*81}s,
which don't support TDS debugging information. There are some tools
which convert a map file to a PDB file, but I didn't got them to convert
it correctly.
I don't know if it's easy to solve or if Borland / DevCo get's enough
information about dump files from Microsoft to support it, but I would
be glad to have direct dump file support in BCB.
Andre
 

Re:Re: Most annoying problems with BCB2006

In article < XXXX@XXXXX.COM >,
Andre Kaufmann < XXXX@XXXXX.COM >wrote:
Quote
I've integrated dump file support into our own applications. They are
written on any errors or on demand, e.g. if the application hangs or got
into an instable state. Our customers will then send the dump file to us
for evaluation and we can evaluate the problem directly in the de{*word*81}.
No need to reproduce the problem.
I would guess that is what the Automated Incident Reports are meant
for. (a feature of BDS 2006) I suspect it is too early to see if they'll
result in a more stable IDE.
--
-David
Quis custodiet custodes ipsos?
 

Re:Re: Most annoying problems with BCB2006

J Alexander wrote:
Quote
"David Dean" < XXXX@XXXXX.COM >wrote in message
news: XXXX@XXXXX.COM ...

>In article <449d67e2$ XXXX@XXXXX.COM >,
>"Dennis Jones" < XXXX@XXXXX.COM >wrote:
>
>
>>There is *no way* Borland could
>>possibly use BCB for anything significant without encountering the same
>>problems and bugs the rest of us do on a daily basis.
>
>Daily basis? If you run into them that often, then you should have
>been able to come up with a reproducible test case. Can you give us an
>example?


On a daily basis I get internal compiler errors. I already used to making random
changes in one header file after which they usually go away. Almost on a daily
basis I have IDE crashes on start or exit. On a daily basis, when I do
debugging, the watches refuse to show the variables because de{*word*81} decided
they were not in the scope or didn't exist.
I'm using BCB6, probably it is the same in BDS. If you are using the
precompiled headers in your project, the compiler generates .csm and
#001. files. After a while if you change something in a header file
which is in the sequence of precompiled headers, the compiler starts to
generate different numbered files like #001, #002. The side effect is
the de{*word*81} cannot show the newly added variables properly. The
solution is to delete this aged .csm and .#00 files, and compile. The
compiler will generate the new .csm and #000 numbered files. The watches
starts to show the variables that you are looking for. Make a bat file
that will delete this .csm files. When you change something in a header
file before the #pragma headerstop run this bat file and compile.
Ho■ca kal
Sabetay
Quote


 

Re:Re: Most annoying problems with BCB2006

David Dean schrieb:
Quote
In article < XXXX@XXXXX.COM >,
Andre Kaufmann < XXXX@XXXXX.COM >wrote:

>I've integrated dump file support into our own applications. They are
>written on any errors or on demand, e.g. if the application hangs or got
>into an instable state. Our customers will then send the dump file to us
>for evaluation and we can evaluate the problem directly in the de{*word*81}.
>No need to reproduce the problem.

I would guess that is what the Automated Incident Reports are meant
for. (a feature of BDS 2006) I suspect it is too early to see if they'll
result in a more stable IDE.
Yes - it's a first step.
IIRC this report is comparable to the madExcept component. I've seen it
once. AFAIK the call stack is sent with this report, a first step but no
replacement for a dump file, which generally holds more information than
only the call stack.
But nevertheless, I'm sure this kind of report would help to track down
the problem, causing the linker access violation.
The only problem is that IIRC the linker only prompts a simple message,
that an access violation has occurred. I suppose the bug report of BDS
2006 is a Delphi component/code and therefore I suppose it cannot be
used for the BCB linker. :-(
Andre
 

Re:Re: Most annoying problems with BCB2006

In article <449e6b87$ XXXX@XXXXX.COM >,
Andre Kaufmann < XXXX@XXXXX.COM >wrote:
Quote
But nevertheless, I'm sure this kind of report would help to track down
the problem, causing the linker access violation.
The only problem is that IIRC the linker only prompts a simple message,
that an access violation has occurred.
It had occurred to me that Internal compiler errors should generate
automated reports, so I wrote a QC report for it, but I don't see any
reason why linker errors couldn't also do it. Why don't you submit a
request for the linker?
Quote
I suppose the bug report of BDS
2006 is a Delphi component/code and therefore I suppose it cannot be
used for the BCB linker. :-(
Maybe not when run from the command line, but why not when run in the
IDE?
--
-David
Quis custodiet custodes ipsos?
 

Re:Re: Most annoying problems with BCB2006

In article < XXXX@XXXXX.COM >,
Sabetay Toros < XXXX@XXXXX.COM >wrote:
Quote
Make a bat file
that will delete this .csm files. When you change something in a header
file before the #pragma headerstop run this bat file and compile.
I've also found that setting a different .csm file for each project
helps to minimize this as well.
--
-David
Quis custodiet custodes ipsos?
 

Re:Re: Most annoying problems with BCB2006

David Dean < XXXX@XXXXX.COM >wrote:
Quote
In article < XXXX@XXXXX.COM >,
Sabetay Toros < XXXX@XXXXX.COM >wrote:

>Make a bat file
>that will delete this .csm files. When you change something in a header
>file before the #pragma headerstop run this bat file and compile.

I've also found that setting a different .csm file for each project
helps to minimize this as well.
That's a must in any case imho - was always so. I dont know why this is still
not defaulted in the ide. Borlanders probably deal just with one project.