During an interesting conversation with a friend this came up
I’d already thought of some things I’d do to fix Xojo given the chance
So here’s my “to do” list (not that it will ever happen)
IF I were ever asked to go back to Xojo it would have VERY different conditions
And at a very different pay rate
I get to be the code Nazi & will check EVERY f*ing check in and they do NOT get merged without my approval
I’d grind them into the habit of “here’s my test cases here’s my code tat passes those cases here’s the check in”
we’re going to PLAN out shit before we change ANYTHING
we WILL do bug fix only release / feature release/ bug fix release
So users dont have to try & figure out what version is “stable” and has NO changes that could f them over
I would build unit testing into the framework. So useful
And there would HAVE to be a way to do UI testing (macOS accessibility etc)
EVERYTHING would have to go through a code review before merge.
I’d add at least 2 or 3 full time QA people
And a dedicated compiler guy
NO vindictiveness in code reviews
The first time anyone says
why the f would you do that ?
thats a firing offence
it better be said like
I have a question about this code. Whats this doing I dont understand. Can you explain it to me ?
SourceTree or whatever is used so that it takes two approvals before we can merge anything.
And when you DO find an issue
add test case that exposes the issue
fix code
test to see code is fixed & nothing else broke
lather rinse & repeat over & over
So your test cases constantly grow
It wont cure anything instantly
But its then a process that can be adjusted & improved
And yes it might cost $ to do
But - it wont happen
I wont get asked back
That I’m 1000000% sure of (I could add several 0’s to that FWIW)
It’s a good list. This is what good software development companies do. Have a process to find mistakes before they make it into production. When mistakes do happen implement fixes so they don’t happen again. Stick to the process no matter what.
I would add:
The product doesn’t get released until development is complete AND it’s gone through a reasonable amount of QA regression testing. If it needs to be delayed a month so be it.
Have one of the QA people, or make it a dedicated position, to be ‘the face’ in the forums. Answering as many questions as possible, and correct misinformation or incomplete, answers from other users. This might even be software evangelist who is constantly doing more/better example projects.
Make some incentives for the beta program users. Find a bug and unit test that finds it and get something (even if it’s just recognition).
Would be nice to see more evidence of that
Like not punting cases down the road
Or having open beta bugs after a release goes out
If you check out the build date on any release you’ll find they are built at most a day or two before shipping
2024r1 was created Mar 25, 2024 and the announcement went out Mar 26
IF there were changes between the last “beta” and that build then they are 100% untested (and this happens)
At most the last change should be from it being set as BETA to FINAL - and thats IT
But, I would bet $ that if I compared the binaries between the two they would have more differences than that
As I said its a PROCESS - like ISO 9000 or anything else
If they followed it quite religiously we’d see the results
And it could be improved incrementally and we should notice the results
The biggest change IMHO would be getting away from features & bug fixes in EVERY release
That leaves people having to make a trade off between new bugs in new features vs bug fixes they need
And thats a shitty trade IMHO
I would add to the list that every ‘X’ number of releases there should also be a complete manual regression test in addition to all of the automated unit and UI tests.
Have the QA people try to break things by going off script and doing things that actual users might do. Just don’t follow the happy path of testing functionality the way it is expected to be used. Especially for new functionality that hasn’t been around long enough to have additional test cases added for found issues.
I truly get where you’re going with the suggestion
But I doubt a full manual test could actually be scripted or even enumerated as to what should be tested
Or completed in a cycle
It would be truly daunting
A good testing set up should not only stress “normal” expected code paths but also “fuzzed” ones for stress tests and bad paths (ui & its underlying supporting code)
This is where automation would be hugely useful to try “good” and “bad” paths
And as this library expands it should get easier & easier to find & fix issues
This is why it won’t happen. In the 20+ years I’ve been working with the company they have not changed their behavior in ANY way.
There were individual developers working for Xojo that tried to make changes and they were always pushed out one way or another. Geoff doesn’t want smarter people than him working on the product - he only wants Yes Men (and women). Considering I work with three former Xojo developers and have developed friendships with other former employees I believe this to be 100% true.
Bottom line is that Geoff is not an intellectually curious person. Otherwise he’d be firing up, and using, other development tools and languages and stealing the best ideas from them. Whatever advantage Xojo had a decade ago is gone and I see nothing coming up that changes their trajectory. Until they prove me wrong this is my belief.
Make one of them a full time problem solving on the forum (Ok, partial time with the tiny community that remains). In the last few days I saw like 4 topics without an answer and at least 2 with the wrong one saying that it cannot be done…
WTF??? potential customers are going to run away inmediatelly with that kind of comunity. Just another thing the genious didn’t think of?
It is a shame that most “mvp” are only good for pollicing but totally useless to actually help people.
Note that NONE of the stated reasons have anything to do with them being helpers on the forum
so there still seems to be no one dedicated to being the face of Xojo on the forums
While I was there I did it because I wanted to
Not because I was required to
I do not think anyone has such a role from the lack of interaction from existing staff
And why they call it a “User support forum” where users can help each other (note is says nothing about the company staff helping you)
Which is not what many seem to understand it might mean
Many seem to expect it is where they can reach the company and get help
ABSOLUTELY agree here
Building the community IS vital for Xojo’s own survival
They need to be active helping their users to create willing unpaid evangelists
They need to cultivate many more third party users & developers to grow the community (like MS did with VB)
Growing an active community is as much their job as writing the software
And such a community will, organically, help grow the user base
