31
Ask a Question / Re: Did I discover Stencyl at the wrong time?
« on: September 26, 2015, 01:56:42 pm »
@ TheIndieStation - I have no doubt that the developers are trying to make things better. I'm just pointing out that the results aren't as good as they probably would have hoped. Specifically, there are too many releases that include pretty severe regressions.
I no longer upgrade Stencyl unless I absolutely have to - it's just gotten too risky.
Now, why is this happening? I can only speculate, but as an outside observer I noticed a few things about Stencyl's release process that probably don't help.
The first thing is that Stencyl doesn't have releases, as such - it has builds. It looks like at some point the developers would decide that what they have is good enough to release, and out the door it goes. Then they continue developing the code, and if users report bugs, they fix those bugs. Then, they would release the current build as the bug fix release.
(Again, I'm just guessing that this is what's happening. Please someone correct me if I'm wrong.)
It's better to handle bug fixes by fixing them in a branch that's rooted in the released version. This way, there can be no risk of introducing regressions along with the bug fixes when all you're trying to do is fix a bug. The bug fixes should be integrated into the main development version, too, but releasing them as patches against an established release is good practice.
There should be a beta period for major releases. Users would be able to download the new releases with the new features, but they'll know that they're more likely to be buggy, so they won't complain as much when they do encounter bugs. It would help Stencyl manage user's expectations.
After the beta period is over, the beta release with bug fixes applied can be released as the official release.
Managed carefully, this kind of release cycle shouldn't really tax the developers' time with too much administration while still setting correct user expectations and making official releases more stable.
I no longer upgrade Stencyl unless I absolutely have to - it's just gotten too risky.
Now, why is this happening? I can only speculate, but as an outside observer I noticed a few things about Stencyl's release process that probably don't help.
The first thing is that Stencyl doesn't have releases, as such - it has builds. It looks like at some point the developers would decide that what they have is good enough to release, and out the door it goes. Then they continue developing the code, and if users report bugs, they fix those bugs. Then, they would release the current build as the bug fix release.
(Again, I'm just guessing that this is what's happening. Please someone correct me if I'm wrong.)
It's better to handle bug fixes by fixing them in a branch that's rooted in the released version. This way, there can be no risk of introducing regressions along with the bug fixes when all you're trying to do is fix a bug. The bug fixes should be integrated into the main development version, too, but releasing them as patches against an established release is good practice.
There should be a beta period for major releases. Users would be able to download the new releases with the new features, but they'll know that they're more likely to be buggy, so they won't complain as much when they do encounter bugs. It would help Stencyl manage user's expectations.
After the beta period is over, the beta release with bug fixes applied can be released as the official release.
Managed carefully, this kind of release cycle shouldn't really tax the developers' time with too much administration while still setting correct user expectations and making official releases more stable.