Qt-interest Archive, March 2002
RE: FileDialog + double click == crash (testcase)
Message 1 in thread
Perhaps they don't provide it because they can't. But then this isn't a
good advertisement for Perforce.
I think the number of bugs I've reported to Trolltech would outnumber
those reported by the casual or not-so-casual developer. For me, with
the lifeblood of my company currently resting on how good Qt is, I feel
particularly powerless to know what other problems there are in even the
released software.
I'd like to see a known list of problems in the released software on a
per-platform basis. That way at least I can make a judgement on whether
or not we release our software with those problems.
However, I think it's all to no avail, which is a real shame. Being
more open would help a little.
-- Paul.
> -----Original Message-----
> From: Paul Robertson [mailto:paul@jprassoc.demon.co.uk]
> Sent: 25 March 2002 11:12
> To: moz@compsoc.man.ac.uk; Arnt Gulbrandsen
> Cc: qt-interest@trolltech.com
> Subject: Re: FileDialog + double click == crash (testcase)
>
>
> >
> > For what it's worth, Trolltech would do much to endear
> develoeprs if
> > they were honest about the bugs in their product. That
> means: a public
> > bug tracker. As it is, the Qt website makes no mention of
> bugs, until
> > they get fixed.
> >
> > This is particularly galling with embarrassing bugs like the one
> > demonstrated above.
> >
> I'd like to see this too.
> I have reported a bug in styles on Windows 98. I'm told that
> it is a known bug and should be fixed in the 3.0.4 release,
> but I can't find out. What other flavours of Windows if, any,
> are affected. Any documented details about the bug Any
> possible workaround I could use in the meantime
> --
> Paul
>
> --
> List archive and information: http://qt-interest.trolltech.com
>
Message 2 in thread
Paul Curtis <plc@rowley.co.uk>
> Perhaps they don't provide it because they can't. But then this isn't a
> good advertisement for Perforce.
Use your brain.
> I'd like to see a known list of problems in the released software on a
> per-platform basis. That way at least I can make a judgement on whether
> or not we release our software with those problems.
I'd like that too. Doesn't mean it's humanly possible to write it.
During the time I was at Trolltech, I probably was the developer who had
the best insight into Qt's various bugs. However, my insight was by no
means sufficient to write a list such as you describe.
I could read a bug report, track down the root cause of it and fix the
bug. That's the easy part. But describing the fixed bug in terms of API
impact? Sorry.
If there's a bug in QWidget::setMumble(), which is called by ten different
classes with a total of 42 publicly documented subclasses, then the bug
may have affected the API of each of those 42 classes, a bug report
generally mentions one of those 42 classes, and the list needs to describe
ten classes properly, maybe a few more if the bug's particularly
insidious, and mention all 42 classes.
Now, you're going to answer: You realize such a list may be infeasible,
but you just want to see the bug reports. Right? But a substantial
percentage of that is support mail from customers. There are NDAs
involved for some of those, and an informal expectation of confidentiality
for others. I heard that well over a thousand bugs were reported in Qt
and Qtopia last year on such terms. And all were fixed.
It would please me greatly if Trolltech now could devote the resources to
build and maintain a public bug database. But I won't hold my breath.
--Arnt
Message 3 in thread
On Mon, Mar 25, 2002 at 02:14:11PM +0100, Arnt Gulbrandsen wrote:
> I could read a bug report, track down the root cause of it and fix the
> bug. That's the easy part. But describing the fixed bug in terms of API
> impact? Sorry.
I can't speak for anyone else but I wasn't asking for this: it would
clearly be infeasible.
> Now, you're going to answer: You realize such a list may be infeasible,
> but you just want to see the bug reports. Right? But a substantial
> percentage of that is support mail from customers. There are NDAs
These arguments hold for most companies involved with open source
projects. Many still maintain at least part of their bug systems in
public. Two good examples would be RedHat and Netscape.
> It would please me greatly if Trolltech now could devote the resources to
> build and maintain a public bug database. But I won't hold my breath.
Sure: it's a wishlist item.
We'd all rather see the bugs just fixed instead :)
regards
john
--
[ signature omitted ]
Message 4 in thread
> -----Original Message-----
> From: owner-qt-interest@trolltech.com
> [mailto:owner-qt-interest@trolltech.com]On Behalf Of Paul Curtis
> Sent: Monday, March 25, 2002 12:37
> To: paul@jprassoc.demon.co.uk; moz@compsoc.man.ac.uk; Arnt Gulbrandsen
> Cc: qt-interest@trolltech.com
> Subject: RE: FileDialog + double click == crash (testcase)
>
>
> Perhaps they don't provide it because they can't. But then this isn't a
> good advertisement for Perforce.
>
> I think the number of bugs I've reported to Trolltech would outnumber
> those reported by the casual or not-so-casual developer. For me, with
> the lifeblood of my company currently resting on how good Qt is, I feel
> particularly powerless to know what other problems there are in even the
> released software.
>
> I'd like to see a known list of problems in the released software on a
> per-platform basis. That way at least I can make a judgement on whether
> or not we release our software with those problems.
There is a list like this at
http://www.trolltech.com/developer/platforms/index.html
I dont think it is complete, and I dont know if this is for the
latest version or not
The page reads:
Qt Platform Notes
Here are the platforms Qt is currently known to run on, with links to
platforms-specific notes, including any known bugs or incompatibilities:
AIX - 4.1 or later.
BSDI/OS - 2.0 or later
DG/UX
FreeBSD - 2.1 or later
HP-UX - 10.20 or later
Irix - 6.x
Linux
Mac OS X
NetBSD
OpenBSD
OS/2 (with XFree86)
QNX
SCO UNIX
Solaris - 2.5.1 or later
Tru64 (Digital UNIX) - 4.0 or later
Windows 95
Windows 98 and Me
Windows NT (4.0 or later), 2000 and XP
XLib on Windows
If you have anything to add to this list or any of the platform- or
compiler-specific pages, please write to qt-bugs@trolltech.com.
So at least Trolltech have intended to have such a list, but
perhaps they dont have a system to keep it up to date.
Erlend
>
> However, I think it's all to no avail, which is a real shame. Being
> more open would help a little.
>
> -- Paul.
>
> > -----Original Message-----
> > From: Paul Robertson [mailto:paul@jprassoc.demon.co.uk]
> > Sent: 25 March 2002 11:12
> > To: moz@compsoc.man.ac.uk; Arnt Gulbrandsen
> > Cc: qt-interest@trolltech.com
> > Subject: Re: FileDialog + double click == crash (testcase)
> >
> >
> > >
> > > For what it's worth, Trolltech would do much to endear
> > develoeprs if
> > > they were honest about the bugs in their product. That
> > means: a public
> > > bug tracker. As it is, the Qt website makes no mention of
> > bugs, until
> > > they get fixed.
> > >
> > > This is particularly galling with embarrassing bugs like the one
> > > demonstrated above.
> > >
> > I'd like to see this too.
> > I have reported a bug in styles on Windows 98. I'm told that
> > it is a known bug and should be fixed in the 3.0.4 release,
> > but I can't find out. What other flavours of Windows if, any,
> > are affected. Any documented details about the bug Any
> > possible workaround I could use in the meantime
> > --
> > Paul
> >
> > --
> > List archive and information: http://qt-interest.trolltech.com
> >
>
> --
> List archive and information: http://qt-interest.trolltech.com
>
>
Message 5 in thread
On Mon, Mar 25, 2002 at 01:45:23PM +0100, Erlend Helmersen wrote:
> There is a list like this at
>
> http://www.trolltech.com/developer/platforms/index.html
Well, this is a very general list, and is nothing like a real bug
tracker. Apparently there are zero bugs on Linux/x86, for example.
regards
john
--
[ signature omitted ]
Message 6 in thread
John Levon <moz@compsoc.man.ac.uk>
> Well, this is a very general list, and is nothing like a real bug
> tracker. Apparently there are zero bugs on Linux/x86, for example.
You mean "zero known bugs". At release time this has usually been the
case.
--Arnt
Message 7 in thread
> John Levon <moz@compsoc.man.ac.uk>
> > Well, this is a very general list, and is nothing like a real bug
> > tracker. Apparently there are zero bugs on Linux/x86, for example.
>
> You mean "zero known bugs". At release time this has usually been the
> case.
>
> --Arnt
So, Trolltech see to it that every known bug is fixed before the next
release?
Really?
--
[ signature omitted ]
Message 8 in thread
> > There is a list like this at
> >
> > http://www.trolltech.com/developer/platforms/index.html
>
> Well, this is a very general list, and is nothing like a real bug
> tracker. Apparently there are zero bugs on Linux/x86, for example.
>
I agree, the list is not complete, but it goes to show that
Trolltech is not hiding all information on problems with the
software...
<zen>It will benefit your well being and peace of mind
if you manage to be greatful for the good things you have
instead of being angry for what you dont have</zen> ;-)
Personally I thing Qt is great, and I find the support team
at Trolltech to be very friendly, competent and patient,
especially once when I had done something outrageously stupid,
but the very competent person at TC helped me through it without
even coming close to calling me st***d.
Have a nice easter holiday.
Erlend
Message 9 in thread
On Mon, Mar 25, 2002 at 02:56:05PM +0100, Erlend Helmersen wrote:
> I agree, the list is not complete, but it goes to show that
> Trolltech is not hiding all information on problems with the
> software...
I wasn't accusing the trolls of hiding information. After all, they
are usually very forthcoming on this list.
> <zen>It will benefit your well being and peace of mind
> if you manage to be greatful for the good things you have
> instead of being angry for what you dont have</zen> ;-)
I'm not angry, and I am grateful. This doesn't mean things couldn't
be better.
regards
john
--
[ signature omitted ]
Message 10 in thread
> > I'd like to see a known list of problems in the released
> software on a
> > per-platform basis. That way at least I can make a judgement on
> > whether or not we release our software with those problems.
>
> There is a list like this at
>
> http://www.trolltech.com/developer/platforms/index.html
>
> I dont think it is complete, and I dont know if this is for
> the latest version or not
Indeed, that's a problem. The list is seriously incomplete. The number
of platform-independent problems in the early releases of Qt 3 was
really quite large. This type of list is not particularly useful.
-- Paul.
Message 11 in thread
> Paul Curtis <plc@rowley.co.uk>
> > Perhaps they don't provide it because they can't. But then
> this isn't
> > a good advertisement for Perforce.
>
> Use your brain.
What does that insult have to do with the supposition that it's
impossible for Trolltech to provide such a list, either because they
don't have the resourcing necessary or the bug tracking framework in
place?
> During the time I was at Trolltech, I probably was the
> developer who had the best insight into Qt's various bugs.
> However, my insight was by no means sufficient to write a
> list such as you describe.
>
> I could read a bug report, track down the root cause of it
> and fix the bug. That's the easy part. But describing the
> fixed bug in terms of API impact? Sorry.
I'm not asking for that. Most bugs are small and trivial. Something
that simply describes what's wrong would be nice.
Now, I've just counted the number of messages I've sent to the Trolls
over the last year. 450 to qt-bugs. Hmm. The number of problems I've
found is quite large. There have been several responses which say:
"Can't reproduce that here, it's probably fixed, go try the latest
snapshot." Ok, so I drag down the latest snapshot and it's still not
fixed. So I say again "Not fixed", and I get the response "Oh, sorry,
yeah, we tested on Linux, but you know, even though you reported it on
Windows XP, we didn't bother to try to repro it on XP, we just thought
we'd try Linux. No way! You're right! You've found a bug, so we'll
fix it now."
This happens. I take the time to send a bug report, with code, saying
which release I'm using and which platform I'm using. The response
"probably fixed" doesn't cut it, especially when it isn't.
There have been a number of problems reported to qt-interest that I know
I've already submitted to Trolltech with example code. Why waste my
time and others time? If there was an available bug list in the latest
release, then I could look up that and NOT BOTHER TAKING THE TIME TO
SUBMIT A BUG REPORT WITH EXAMPLE CODE ONLY TO GO THROUGH THE PAIN OF
BEING TOLD IT'S ALREADY FIXED? Even worse is being told it's fixed when
it isn't.
> If there's a bug in QWidget::setMumble(), which is called by
> ten different classes with a total of 42 publicly documented
> subclasses, then the bug may have affected the API of each of
> those 42 classes, a bug report generally mentions one of
> those 42 classes, and the list needs to describe ten classes
> properly, maybe a few more if the bug's particularly
> insidious, and mention all 42 classes.
Nah, you just describe what's wrong in QWidget::setMumble. Most users
will understand what the implications are. I'm not asking for an impact
assessment of this, just a list of things that are a out of kilter.
> Now, you're going to answer: You realize such a list may be
> infeasible, but you just want to see the bug reports. Right?
No, I don't want to see bug reports.
> But a substantial percentage of that is support mail from
> customers.
I'm one of them.
> There are NDAs involved for some of those, and an
> informal expectation of confidentiality for others. I heard
> that well over a thousand bugs were reported in Qt and Qtopia
> last year on such terms. And all were fixed.
Thousands? Gee. Seems that this software is choc full of bugs. I
don't give a damn about the reports, I give a damn about my time though.
I pay Trolltech for the product and for support. It would be nice if
they could take the time to collect a small amount of text together a la
GNU change logs, that would mean I don't waste my time reporting things
that are already fixed.
> It would please me greatly if Trolltech now could devote the
> resources to build and maintain a public bug database. But I
> won't hold my breath.
So, that seems to support my view that Trolltech can't provide a known
list of fixes or bugs.
-- Paul.
Message 12 in thread
Since there have been some discussion over the fact that there is no known
bugs lists the past few days, I'm here to make an official response on the
current situation with this in Trolltech.
We appreciate your views and we are working on providing something of this
nature for the future. Internal development is taking place for this, and
we apologise for any inconvenience caused by the lack of a such system.
Please bear with us.
Thank you for your patience.
Andy
------------------------------------------------
Mailing Lists Administrator Fax: +47 21604801
Trolltech AS, Waldemar Thranes gt. 98, N-0175 Oslo, Norway
Message 13 in thread
On Mon, Mar 25, 2002 at 05:04:54PM +0100, Andy Shaw wrote:
> nature for the future. Internal development is taking place for this, and
> we apologise for any inconvenience caused by the lack of a such system.
Great stuff. Thanks !
regards
john
--
[ signature omitted ]
Message 14 in thread
> John Levon <moz@compsoc.man.ac.uk>
> > Well, this is a very general list, and is nothing like a real bug
> > tracker. Apparently there are zero bugs on Linux/x86, for example.
>
> You mean "zero known bugs". At release time this has usually
> been the case.
Actually this is not true. There are a number of serious, known bugs
that have not been fixed since Qt 3.0.0.
-- Paul.
