Recent builds

I've removed the two recent build topics, simply because they've been nothing but trouble.

There's just been horrendous miscommunication on everything, and it's for the better users don't have to deal with the problems till they're all entirely sorted.

For now, use 0.98.7, we'll have something sorted soon.

I don't mean to criticize, but perhaps it should be best to treat the next (or any) new version with the thought: don't implement new features, unless 90% of the known (significant) bugs are fixed.

Its not that the new features are causing bugs its just sometimes bug fixes can lead to unexpected bugs other places in the program. Plus there is no way to test every way possible that the code could be used someone will always use it in an unexpected way That is how you find and fix the bugs which is wha they have been trying to do. I usually wait a little while to see how the build is before I update my caps so it dose not cause problems for me.

Plus all the devs Im sure do their best to fix as many bugs as possible before the next build but some of them are really hard to understand when people submit them and some of them are not bugs at all.

I personally say now is the time to release any and all plugins while you are still in beta and have a good sized group of people to test with. What is the point of racing to 1.0 if all the features you want are not going to be in it?

I think that Stable and unstable builds will defiantly pay off in the end.

Separate stable and unstable builds will help, although most people should be (and probably are) aware that bugs are going to appear when changes to code are made.

Software has always been the same.

Another idea you might consider is a beta test area for those who are prepared to put in some time to test each build to it's limits for a week or so, so that fixes can be made and a stable build released from that.

As a former beta tester for commercial software, (and a tester of my own software for 20+ years), I'd be glad to be part of that.

In fact, I'd consider it a payback for all the work you've put into Construct already.

Either way, well done on this last build.

Sponge.

The last build problems were silly and a result of testing which wasn't good enough, we'll ensure it doesn't happen again, and in any case, we're just about done changing things which will mess up .

....and in any case, we're just about done changing things which will mess up .

We'll hold you to that.

As deadeye pointed out in the chat, aren't forum releases classed as unstable and the front page releases stable?

Its not that the new features are causing bugs (...)

Plus there is no way to test every way possible that the code could be used someone will always use it in an unexpected. (...)

I am replying you, Aeal5566, to clarify that I didn't have the impression that new features necessary cause the (newly found) bugs.

My post was about taking small steps and thus focus on as less things (first known bugs, then new features) as possible.

Also it's nice you mention and clearify that there is no way to test every way possible, but my post was about bugs in the bug-tracker, and its description, given by the submitter. I got the impression you thought I meant -all bugs- and -all code-.

But, that was the idea of my first post.

Now that I discovered the concept of a stable and unstable build, I think the developers are doing fine, also concidering Rich mentioning the last build included "silly mistakes". So I'm thinking of withdrawing my first post.

We are in recent builds doing less feature work and more bug fixes, but we have some features we literally have to get through before 1.0. If we did them all at once, it'd introduce more bugs than doing it bit by bit.

Obviously users want a stable Construct, but compared to 12 months ago, we're miles ahead in stability, and features, so it's working. We just need to do that little bit more testing to ensure obvious mistakes don't occur, and everything's gravy.

There's a spot on the home page for nightly builds:

http://www.scirra.com/nightly.php

Couldn't you just use the Dropbox service, and place the untested builds there?

All you would need is one link, and just rename the file to something like Construct-latest-untested.exe?

Heck you could even have a Construct group folder to put it in, that way you'd always have the bleeding edge build.

We could do, but the builds we release generally have been tested, just not to the extent which hundreds of people using it will show, there'll still be some issues - even if very small - in most builds till 1.0.

I like the idea of it personally, so I'll probably do it.

i bet that when you finish 1.0 , you will build 2.0 from( or mostly from) scratch, because i'm preatty shure that while you building up to 1.0 version, you think "that this part of softwer could be done difrent, but it's to late now"?

i bet that when you finish 1.0 , you will build 2.0 from( or mostly from) scratch, because i'm preatty shure that while you building up to 1.0 version, you think "that this part of softwer could be done difrent, but it's to late now"?

It's sod's law, and you always think of "better" ways of doing it when it's after the fact.

I'm sure I read on one of the main pages that they intend to drop the 3rd party GUI after version 1, and utilise the normal Windows API.

I'd welcome that.

Sponge.

That's right, we want to use MFC 10 (when it's released) and ditch the current UI library that we have to pay for out of donations at the moment.