[Scummvm-devel] Windows console

Travis Howell kirben at optusnet.com.au
Thu Oct 28 06:44:34 CEST 2010


On 27/10/2010 8:14 AM, Max Horn wrote:
> Am 26.10.2010 um 15:48 schrieb Travis Howell:
>> On 26/10/2010 7:56 PM, Max Horn wrote:
>>> However, I do stand to the part where I say that this has been brought up before (on tracker items, on forum discussion, on IRC, and in real life conversations with users), and that there was never a proper discussion on this. At least not that I recall, but if I forget something, please be so kind and refresh my memory.
>>>
>>> As such, I still think it's important that we discuss this thoroughly, and without letting personal feelings get into this (which is why I feel doubly bad for violating just that with my mail earlier).
>>>
>>>
>>> So far, we exchanged (and still are exchanging) technical points, plus not-well based claims about what our user may or may not want. The latter is something we don't really have any good data on. I therefore think that we should run a poll, and ask people about what they want. The poll should be worded carefully, as to not unintentionally bias people one way or another. Here is a suggestion for what we could ask; let's see if we can shape this into something we can all agree on, then we can put the poll online (e.g. in the forums or using LimeSurvey, other suggestions welcome):
>>
>> No, any internet poll would only cover a small amount of users.
>> Specifically users who have online access, and actively monitor the
>> ScummVM web site. Please don't turn this into another pointless online
>> petition.
>
> Any opinion poll always only reaches a fraction of all people, that's an inherent property of opinion polls. But if we do it right, I don't see why we shouldn't get a couple hundred responses. And while those of course will not tell us what every single ScummVM user likes or dislikes, they can give us a quite good estimation.

But an internet poll is much worse in several ways:
An internet poll isn't accurate, whether confirmed via email or IP 
address. It is too easy for a single person to increases votes for one 
option.

An internet poll can't even guarantee a random sample of people. A 
person strongly interested in one option, can simply gather a group of 
people for that option, and the poll is biased.

An internet poll completely excludes offline users, where the console 
output could provide the most benefits. Since offline users have no 
access to other means of help, and have to rely solely on the 
information provided by ScummVM.

> Anyway, so: We are discussing an issue that some people (well, me at least) think is bad for users, which you dispute, but at the same time you refuse to ask those very users based on the argument that we won't be able to reach every of them?
>
> So... I did at least talk to some users about what they like or dislike; admittedly far to few to give a representative picture. But you seem refuse to even talk to them, but still think you know what's best for them? Hm.

Again, you basing on a flawed sample of a few people, and still have not 
provided proof that strong change is wanted. I could do a similar flawed 
sample of a few local people using ScummVM, only to get similar results 
for another option (i.e. simply do care either way).

I already stated the lack of requests, over the many years that ScummVM 
has been available for the Windows platform. With the few requests been 
for an option added to disable the console window, rather than changing 
ScummVM to no console window by default.

>> Besides any poll of this type, could set a dangerous precedent for
>> ScummVM. Since we never add features or games, based on requests.
>
> It was actually a meant as a gesture of good will towards you. Because it currently looks as if we are not able to come up with a constructive consensus, and then the only alternatives I see are either some kind of poll; or else it'll be the team leaders who'll make the final call, and I'd prefer to avoid that.
>
> Anyway, I still have not completely given up on the faint hope for an agreement. You did raise a good point regarding many engines using warning() to send messages to the users, e.g. with report about save failures (sometimes these will also be shown with a GUI dialog, but usually with less detail than the warnings provide). This is a typical disease of software; developers tend to be very lazy when it comes to writing good error handling (I don't claim to be an exception ;). This is bad for the user. (Note: I do not agree that this is a justification for showing the console by default.). Anyway, this is something we should address. I wonder what you would say if we were to "fix" all these warnings to be properly reported to the user, without relying on a DOS console (which, by the way, is not usually visible on every other port, i.e. all those poor users have to get along without it already).

Again we have the issue of a team leader, over riding the maintainer of 
a specific port, without good reasoning.

I'm basing my opinion, more on what would be best for the project. 
Windows continues to be one of our main platforms, and we are relying 
even more on release versions for feedback, due to reduced testing 
during release cycles. The console can provide useful information for 
feedback on bugs/issues, and even useful information for users (no 
matter how un/experienced) at times.

Actually this whole discussion has focused on changing the Windows 
builds to no console by default, rather than previously showing a 
console window by default. While the few requests in the past, have been 
only to add an manual option for no console window. Which I would be 
more agreeable to, even if the implementation can't be perfect.

I added a patch to the feature request item on the tracker, which offers 
optional disabling of the console window in Windows builds, but doesn't 
change the default behavior of opening a console window in Windows builds.




More information about the Scummvm-devel mailing list