Alpha lowers the expected level of polish and stability - it doesn't make every regression equally acceptable. It is a reasonable criticism even when released to the alpha channel.
Further, I believe it's reasonable for users' expectations to be influenced by what the project itself communicates about the development process.
People sometimes.
I have experienced many (relatively minor but some annoying) bugs in GrapheneOS the past few years, without complaint to the developers. And have submitted bug reports and logs. I'm "in the trenches", not just slinging mud from the outside. If I were in a better financial situation right now, I would be donating too.
It's doubly disappointing because just 4 days ago the project wrote [1], on using AI for the new Messaging app:
Why? Using an LLM as an additional reviewer seems like a good use of LLMs? Personally I have found LLMs very useful for that and they catch issues that other experienced programmers do not always find. It’s not like they are vibecoding a messages app.
Comments
[flagged]
The notification the Messaging app produces should have the option to copy the one time code directly.
It's all 'links' (phone numbers, URLs etc) that are broken. Not just one time codes.
You're using software from an alpha channel.
People sometimes.
Alpha lowers the expected level of polish and stability - it doesn't make every regression equally acceptable. It is a reasonable criticism even when released to the alpha channel.
Further, I believe it's reasonable for users' expectations to be influenced by what the project itself communicates about the development process.
I have experienced many (relatively minor but some annoying) bugs in GrapheneOS the past few years, without complaint to the developers. And have submitted bug reports and logs. I'm "in the trenches", not just slinging mud from the outside. If I were in a better financial situation right now, I would be donating too.
Conflicts with the common definitions of alpha versions I'm familiar with but definitions differ, of course.
Glad you're not among the gOS haters though!
Pointing out regressions wouldn't make someone a hater. That would be a nonsense accusation and a childish conclusion to draw.
This is somewhat true. I expect more regressions from released commercial software.
I suspect there are fewer regressions per release on graphenes alpha messenger than there are in googles message's app.
You make a good point: as a project becomes more commercialised, you can expect regressions like this.
It's doubly disappointing because just 4 days ago the project wrote [1], on using AI for the new Messaging app:
Why? Using an LLM as an additional reviewer seems like a good use of LLMs? Personally I have found LLMs very useful for that and they catch issues that other experienced programmers do not always find. It’s not like they are vibecoding a messages app.
What do you mean "why"? For the exact reason he stated in his first paragraph - a serious regression impacting his workflow.
... which seems to have nothing to do with using an LLM to do extra review.