Is there a current tracking list for issues needing to be resolved in order to come out of beta?
It used to be tracked on a Github milestone, but the scope kept creeping so they don’t appear to be maintained.
Going through the issues to check if they are appropriately labeled could also help a bit, but it may be a rather time and energy consuming task. And it requires some knowledge of the internals.
I think so.
It’s also tough because there isn’t strictly a definition of done. So any issue could be put in there.
Perhaps there needs to be a good definition of the expected features for coming out of beta otherwise it never will since it is a moving target.
Key things to me would be:
- no backup data corruption, including application crashes during backup (though in-progress backup might be invalid/lost?)
- listen for a system shutdown and stop the backup cleanly (Windows Updates might reboot the machine at night)
- correct handling of all symlinks
- correct handling of long paths (windows)
- complete language translations
- use of subfolders for backup destination (to me this seems pretty important and for some others since large backups are would otherwise be limited)
Would some user polls be useful to help narrow features/fixes to get out of beta?
I think we should put in the effort to use the GitHub tools (labels, projects, milestones, etc.) to our advantage. In my opinion, management of important issues, tasks, and release milestones is far superior in GitHub compared to this forum. For example, I think topics such as Enhancements when out of beta might benefit from residing in GitHub.
As an example, I think the Brave project is using these tools in a rather effective manner. Of course, this is highly dependent on the availability of developer time.
Regarding @BlueBlock’s list, I think that it might be a little too ambitious. I agree that fixing the data corruption issues should be the number one priority (the difficulty here is the lack of easily reproducible test cases). In my opinion, completing the language translations and use of subfolders are less important. Keep in mind that there have been a lot of fixes since the last beta release that users should be getting.
Sounds good to use the github features. Is there a current milestone set for coming out of beta?
The subfolders probably are less needed since it only impacts possibly a few backends but OneDrive being a big one. I have is basically done except for updating all backends… so I think I could make subfolders available only to the backends modified for it. Then enable additional backends as they get updated to handle subfolders.
There is now. Since the upcoming version numbers are a bit unpredictable, I named this milestone “Upcoming beta”. Ideally, the changes in this milestone would first be made available via canary and experimental releases before appearing in a beta. I think we can attempt to keep track of this via labels (e.g.,
I’m new to GitHub milestones, but this plan seems reasonable. I’ve long wanted a lightweight way to focus.
Critical issues with the current canary [18.104.22.168] had a different method, and neither is a fancy tracker tool…
Any time there’s tracking, even just GitHub Issues, there’s a certain amount of administration so let me ask whether initial use starts with self-service honor-system as to what’s added? I hope justification is supplied unless it’s hugely obvious. This would be somewhat similar to promoting a feature request except maybe a feature request would have to work harder if the milestone is already full of critical bugs that need attention.
CheckingErrorsForIssue1400 and FoundIssue1400Error test case, analysis, and proposal #3868 is one I’m pointing to as a possible example of how to justify making the Upcoming beta milestone. It’s much-reported (statistics supplied) and reproducible (steps supplied). Those two make it potentially actionable, but for any near (and I hope it’s near) milestone, the bar may be higher. This one is analyzed to code, and has fix PoC. What is doesn’t have is an assignment (which Brave tries to use). I propose that as a “bonus” for an “Add”.
I also propose that regressions from previous Beta found in Canary or Experimental get on list more easily although not automatically. For a given area, quality should increase, even if initial ship is less than perfect.
Regarding what gets added to a milestone, in my experience it’s typically up to the developer(s) who maintain the release. The community can certainly lobby for a particular feature/fix, but ultimately it depends on the availability of developer time and their expectations of what can realistically get done.
I do think that more frequent releases would be helpful (and reduce the urge to keep adding “one more thing”). In fact, for this current beta release cycle (which will hopefully end soon), I would prefer that we begin to focus only on testing and bug fixes, and no longer include any more new features. The more things we try to include, the more testing is needed and the longer the delay. As you already know, there are a lot of changes in the queue that beta users have been waiting a long time for.
@BlueBlock, I just re-read the title of this thread. Are you asking what needs to happen to create a “stable” release? If so, then many of my comments here might be out of place. Some of my comments were with regards to the next beta release, and not a “stable” release. Sorry for any confusion that this may have caused.
I’m thinking of both the next release and the next beta. It would be great to have a narrow list of items for the next beta. And then a set of features targeted for coming out of beta.
It would be great to have frequent betas going out. Weekly for the small bug fixes or even multiple times a week. And then monthly? for bigger fixes or features.
Having such a large span of time between what users are running and the code base can make it difficult to identify problems.
I’m new to the github features for releases. I’m not familiar with how we move issues/PR’s to target different releases like canary. I just need to get an understanding of how to see what issues/PR’s are targeting canary etc. and then how PR’s can get moved between different releases or milestones. I just need to do some youtube education for github features LOL. We must be able to work on long-term features easily… maybe just more branches with a monthly targeted release. And we move PR’s around as we see fit? Not sure.
I wrote a plan on releases a really long time ago here: Discussion: release cycle
But the primary reason it’s not easy to implement any kind of “filtered” release seems to be the difficulty of removing a pull request/feature that was already merged.
Due to this our flow now is strictly “once everything in master is ready then we can upgrade it to next stage”. Which of course creeps on forever.
I think Kenneth has been hesitant to push any more releases exactly because of that overhead and uncertainty about the quality of recent changes.
On the flipside i try to make sure all pull requests have feedback or is merged within a short time to avoid discouraging contributors. A short turnaround time between opening the PR and seeing it in canary is in my mind the only way to keep people involved in the process.
I have to look how other projects handle this. I think your release cycle plan looks good.
I’d seen that plan before, but lost track of where it was. It touches on some of the same points I made in:
Maybe I should just cut-and-paste my bullet list here? Though user-inspired, it’s only developer-fixable…
I’d love to hear what @BlueBlock finds out, and thanks for raising the issue, even if it’s wandering a bit.
I know we’re going on various topics here but it seems a good place to address some common issues.
To get the next build out, besides any WIP PR’s, it seems like if we can wrap-up the existing PR’s and put a build out? Would we put out a canary build first?
p.s. I am wanting be on .net 462 because I feel it is important to get there for reasons I’ve outlined before. I’ve think I’ve addressed concerns about user impact. So i’m hoping concerns have been addressed and that PR and move. I’m not trying to drag the conversation here, if needed we can take it to the .net 462 PR.
Getting there, but it’s making the scheduling look worse. I took a break yesterday, but just added more. There’s been a wish to get broader developer input on it, so anybody who hasn’t been asked, feel free. Current question is on user benefits of .NET 4.6.2 versus 4.5.2, and proper prep for 4.6.2 if it goes out. “Important” does not directly translate to “Important for users to have right now”, or does it? Discuss…
On the linux mono side, users should at least be at mono v5 if the user followed the installation instructions.
On the Windows side for the user there is little difference between 452 and 462.
I’m not sure how it impacts scheduling if we’re talking about going to 452 versus 462. They bot would take an equal, actually identical, amount of time. And I’m not sure I really understand what impact to scheduling you are seeing.
Duplicati 2 came out in 2014. The mono directions might be from March 2018, per What about a manual?
Based on that, making users maybe advance mono by 27 months (some risk/work) seems like a net loss.
.NET 4.5.2 target might be safe enough to run on old mono without taking time to go heavy-warnings route. Agree that canary testing lots of updated libs would take awhile either way, so still hoping for basic
Detailed in “Update framework to 462” PR. Biggest delay is if the warn-before-requiring-it plan is deployed.
And so was determined as the proper installation of Duplicati on linux.
You seem to be relying on outliers like a user on mono 4.6
How about require users to have mono v5 from 18 months ago? That sure seems like a good compromise doesn’t?
Again, this is all in order to support mono 4.6 and 4.8 users who should be on v5 if not v6 per the installation docs, let alone proper system maintenance, bug fixes, getting to TLS 1.2 etc.
The aftp issue is fixed.
That related to staying on 4.5. Moving to 4.5.2 would have the same requirement as moving to 4.6.2
Some info development cycle that I’ve been reviewing. I’m getting used to git in a group as I’ve dealt almost entirely with enterprise environments.
It looks like there are two trains of thought.
One being trunk-based, but I think we can toss this immediately for open source as it does not provide code review and hence low security.
The second is likely what we might want to follow and it is Git Flow.
GitHub has a nice explanation here at time 3:07: