Re: [Suggestion] GNUstep-test for quality control

[Ian Jones]

Hi All,

On 2003-10-16 08:26:02 +0000 Dennis Leeuw <dleeuw@made-it.com> wrote:

> Hi Stefan,
>
> An important note: the Users of GNUstep are developers. GNUstep is a 
> development environment. That's the problem.
This is not strictly true, I would not consider myself a developer. I have 
dabbled a couple of times with gnustep, made a service, bundle for talksoup and 
an odd patch here and there.
I am mostly a user of gnustep applications, I'm not sure how many others fit in 
this category.

> So to get more Developers to develop programs with GNUstep you have to 
> attracked them. One way to attracked new developers might be to provide next 
> to a great API a great testing environment.
I think the testing environment will be of use only to GNUstep developers. 
Users or app developers probably won't touch them.

> This way you can tell a developer when he runs into problems if it is his 
> mistake (his app) or the GNUstep API. That might prevent them from running 
> away from GNUstep and just file a bug report instead.
I  have ran into problems when developing applications because of bugs that I 
thought were in my own code that later turned out to be in GNUstep itself. I 
gave up on what I was doing at the time because I just couldn't get it to work. 
This is one guaranteed way to lose new developers. When things don't work as 
you think they should and as they are documented it becomes very frustrating.

> And automated mail-bug-script would be ideal.
Savannah works just fine, but maybe it could be publicised a bit more, for 
instance I never knew about it until a couple of months ago. The first and only 
bug I posted on there is the one about the recursive thread locking that seemed 
to spark this whole conversation.
I think you'll find people don't mind spending a couple of minutes to make a 
bug report as long as they feel something is going to be achieved by it. I got 
a little frustrated because after adding the bug to savannah along with a 
backtrace it it was dismissed by one developer before Alex Malmberg took a look 
at it and managed to come up with a fix which stopped the recursive thread 
locking I was experiencing.

I'm also not against testing frameworks and the like, but I don't think they 
are any substitute for people using and developing with GNUstep, it seems like 
the application developers are the ones who are suffering most. They discover 
bugs in GNUstep and either have to fix them themselves and get a gnustep 
developer to apply their fixes or wait for a developer to get around to looking 
into the problem they are experiencing. Often these problems make the 
application and or the developer of the application look bad when in fact it is 
not their fault their application is crashing or exhibiting strange behaviour.
In the past I have wondered how many people who develop gnustep applications or 
GNUstep actually use these applications in every day life. I have thought this 
to myself on a number of occasions when things just don't work as they should 
or fail to work at all. Often it has felt like gnustep apps are nothing more 
than something to play with on a sunday afternoon. I know this isn't the case 
for all applications, but it is the feeling given by some.
I use quite a number of gnustep applications now and look forward to there 
being more, but I certainly value quality above quantity, it really annoys me 
having to keep a duplicate non-gnustep application around for every gnustep app 
I have for when either the application stops working either becasue someone 
missed something silly in their code or someone has altered gnustep cvs and not 
bothered to test that applications still compile and work with the changes.

To fix this, well, I don't know, I'll leave it to people better able to make 
these decisions, I just give you my experiences as a user and occasional 
developer.

> Hope this helps,
>
> Dennis

Regards,

Ian