FC5T2 - not ready for prime time.

[John Summerfied]

Peter Jones wrote:
> On Sun, 2006-01-22 at 09:13 +0800, John Summerfied wrote:
> 
>>Jesse Keating wrote:
>>
>>
>>>So while the bug(s) may seem obvious to you, unless your exact hardware
>>>is sitting in my cube it isn't going to seem very obvious to me.  I am a
>>>bit hurt at the implication that we don't even bother to test our
>>>release before we send it out the door.  
>>
>>You shouldn't be. Have you forgotten the users' perspective so soon? If 
>>one puts as much effort into installing it as the OP appears to, then 
>>the conclusion that there's been unsufficient testing is a reasonable 
>>one for a user to make.
> 
> 
> I've probably done (well, at least started) 300 installs so far this

Um, Peter, I was addressing my remarks to Jesse.

I worked on IBM's PL/1 compiler for a time, and my major task was 
testing the compile against a supply of around 11,000 testcases, some of 
which dated back to bugs reported in the early 70s - 20 years before. I 
know a little about testing.

> month, if not more.  Like it or not, the way humans work largely
> involves acclimating ourselves to patterns.  So for some bugs, which I'm
> particularly *searching* for, I'm going to find them.  But the sheer
> number of installs I do is going to render me blind to many things.  The
> same holds true for many developers.
> 
> I could do 1000 installs in a month, and it wouldn't make things
> *better*.  This is why we have test releases -- so people who don't do
> 10+ installs per day will see it and report what *they* see, rather than
> what those of us who are unquestionably acclimated to both the code and
> the behavior see.

OTOH I have reported a memory leak in Mozilla (to Mozilla.org I think, 
but it could have been Debian). There is something about the way I use 
Mozilla that causes it to consume all of available memory and to not let 
it go until it ends. Others have reported the same kind of probkem, and 
on a variety of platforms including Windoes (no, it wouldn't have been 
Debian).

The bug was closed because the developers couldn't recreate the problem, 
and since I cannot run Mozilla without seeing the problem I cannot avoid 
the problem any more than they could create it.

I don't know whether the problem's fixed now; I'm running with a Gb of 
RAM and my usage patterns may have changed.


The point is not how much testing you're doing; more important is what 
you're testing, and it seems likely that you are not testing what the OP 
is trying to do.

If he and you can get together over a VNC session, then maybe he can 
show you.


> 
>>Hinting that BZ is a good idea is fine, but I for one find it a royal 
>>pain to use, and nobody's found time to write an email (or similar) 
>>interface that Mike Harris agreed was a fine idea before he joined RH.
> 
> 
> Long ago I worked in support, when you could email in a support query
> *with* authentication to prove you were a paying customer.  That's a
> significantly higher of a barrier to entry than the panacea which you're
> suggesting.  I really wish you could still email support questions and
> hope to get a reasonably answer.  I totally disagree with the idea that
> this feature is even anything remotely like a good idea in a bug
> reporting system.  All that it will accomplish is more reports.  Not
> more legitimate bugs actually being reported.
> 
> 
>>More important is asking for the specifics of what went wrong, and 
>>asking in _this_ forum is way more likely to get related input from 
>>others than you could hope to get with bugzilla.
> 
> 
> Have we forgotten the start of this thread?  The original post included
> this text:
> 
> 
>>Probably 8 out of 10 times the installer died while I was picking
>>packages to install.  Each time it would pop up a message box saying
>>something to the effect 'this is a bug' and describing how to report
>>it.

Evidently he's doing something unlike what you're testing, Peter.
> 
> 
> Not exactly specifics, and the instructions (which clearly were
> presented) have not been followed.  Nor have details emerged in response
> to our queries for descriptions of the particular configuration in
> question.

I've not seen his problem; I've not yet tried to install t2, but I'm 
sure I saw something in its documentation telling testers to report 
problems here. That's what he did.






-- 

Cheers
John

-- spambait
1aaaaaaa at computerdatasafe.com.au  Z1aaaaaaa at computerdatasafe.com.au
Tourist pics http://portgeographe.environmentaldisasters.cds.merseine.nu/

do not reply off-list