I just noticed that the GSOC 2021 Ideas page was in need of a bit of a clean-up, so I’m taking the opportunity also to ask if anything needs to be added/deleted to reflect recent information (i.e. possible mentors who might no longer be available, ideas that are redundant because they have already been implemented by developers etc.). I recall @PulkoMandy saying on the mailing lists too that GSOC will be a bit different in that it’ll allow smaller tasks so some ideas may need to be split into smaller portions, and the requirements are being changed so we may see more people participating. Some work has already been done, but the page may need more work (i.e. adding new tasks or updating existing ones) not just for this year’s GSOC but also future GSOCs.
If you feel anything needs to be added/changed feel free to suggest below or you can edit the page yourself on Github.
I would vote to remove the item for a virtualbox port, we already have qemu and energy could be spend improving that.
Add to that the compatibility problem virtualbox frequently has and the scummy stuff oracle does (e.g with the guest additions) i really don’t think it is even a good idea to port.
Just my 2 cents ;)
(edit: why does discuss constantly replace emotes by emojis, ugh)
One more thing - the ideas page also has some “hidden” ideas (hidden in HTML comments) for various reasons - these also need to be looked at and either made visible or removed if necessary. Since these are hidden in the HTML though you may need to look at it through Github:
I keep the page mostly up to date and review the “hidden” ideas. A lot are disabled because we have no one in the mentor team who can mentor them.
In general the ideas that are there are up to date.
Also, for this year it is slightly too late to worry about this - we already passed the review of the ideas list from Google. But it’s useful to keep it up to date for the next years as well.
Things to remember:
Only add ideas there if you are willing to mentor them. It’s useless (and not nice to the students) to have ideas exposed there, and then have to reject the students because no one can mentor the ideas.
We will accept as mentors only people with commit access to Haiku, or former GSoC students. We have tried to make exception in the past and it did not work (with the mentor suddenly disappearing in the middle of the project, for example)
Remember that a lot of the ideas listed there are not the high-priority things the Haiku team plans to work on. The reason is, we often can’t put these new things in the hands of students who don’t yet know Haiku very well. As a result, we have to go with things that are a bit more “side project” and a bit less critical.
The main thing we need currently for GSoC is not ideas, it’s mentors. There are only 5 persons registered as mentors this year (including the two org admins). Which means we will only be able to accept 2 or 3 students.
I agree with improving QEMU on Haiku and even make KVM acceleration available. Efforts on these two would be a better assingment than porting Virtualbox, imho.
Agreeing with all of the above, but also adding in VirtualBox Guest Additions for Haiku as another suggestion too. Having better integration with VBox can help improve the experience of Haiku users using it. Note that this is not porting VirtualBox itself to Haiku, but rather improving the Guest Additions that used to exist.
@3dEyes compiled the latest VirtualBox Guest Additions a few months back however they KDL Haiku on boot. Once he could fix that we could re-enable the package.
Does that cover Guest Additions? And even if it does, I don’t think that the Inc. would be affected by it since work on Haiku’s Guest Additions is done by independent community members.
On the other hand, I think that Haiku should be more accommodating to people who want to try it out using whatever virtualization software they know.
Out of all the cross-platform options that I’ve suggested to people curious about Linux and virtualization in general, most preferred either VMWare or VBox due to the ease of setting up VMs. There is already VMWare integration in HaikuDepot, so having VBox integration improved and back on there too will improve the experience of those trying out Haiku in VBox. That’s good for adoption of the OS.
And for the record, I am far from being a fan of Oracle as a company. But I have to recognize that there are a contingent of potential users that prefer VBox for various reasons. Also, unlike VMWare Player or Workstation VirtualBox is mostly open-source with the notable exception of Oracle’s extension pack.
Oracle is actively hostile towards their users, vm software or no vm software, why exactly should we support them? For pretty much every OS virtualbox runs on you have better or even native hypervisors instead, which would make much more sense to target if we want to improve our vm experience.
Windows? hyper-v
linux? qemu
FreeBSD? bhyve
illumos? bhyve
macos? xhyve/hypervisor
netbsd? nvmm
With all these options that we partially support, why focus on pretty much the worst of all of them?
We could spend energy into hyper-v guest additions (microsoft has a spec on this, we can ‘just’ build a driver, no third oarty stuff neccesary), or for example activate qemu-ga, improve our virtio support. etc
Performance isn’t everything, or at least isn’t something that all VM users will prioritize too highly. For more casual users or those who want to set up a VM quickly, ease-of-use takes priority over performance. In this regard, VirtualBox and VMWare Player/Workstation do very well among cross-platform virtualization software. Haiku already has VMWare integration. VBox Guest Additions needs some fixing up, which is building upon pre-existing work instead of writing new code like a driver.
We both agree on not really liking Oracle much at all, but I’m also thinking about potential users as well. In particular, users who may be curious about Haiku but don’t know enough about using Hyper-V, QEMU, Bhyve, etc. to quickly set it up. I don’t think it’s a bad idea at all for Haiku to be a great experience on any virtualization software, as that can drive adoption.
It’s really not a good idea at all to tell those who are curious about Haiku to just learn this hypervisor or that VM software if they already know or prefer one. It is already quite difficult to even get people to look at a non-Linux OS, much less try it. At the very least if someone is intrigued enough to try Haiku in a VM software of their choice, make their experience a good (if not great) one so that the chances of them using it more become higher.
In any case, the suggestion is out there already. I am not interested in taking this argument any further. Would rather see more ideas for GSoC put forward, instead of yet another heated internet argument stacked with walls of text.
This is about the proprietary extension pack you can install in the Virtualbox Host, not about the guest additions which are open source. Doesn’t make Oracle’s actions less despicable but still an important little detail.
As a student this year who is very interested in applying to work with Haiku over the summer, I’m wondering if there a list of the active mentors for 2021?
I’m specifically interested in porting Golang to Haiku (one of the listed ideas) as I’ve already had some experience with HaikuPorts and am interested in learning more about compilers. I can appreciate that those listed “possible mentors” aren’t available this year, would this year’s mentors be potentially averse to such a project?
All project ideas are valid and we will probably manage to find a mentor. Not all of us know a lot about Go but we will be able to help with the Haiku side of things at least. If you are working on an idea like this it may also be a good idea to get in contact with the Go community and see if they would help with mentoring (we occasionally did set up projects with multiple organizations involved like this)
I have been considering mentoring GSoC again this year after a very long hiatus and mentoring a Go port might be right up my alley. I’ve been programming Go for many years and it is my main language at work. I’m not an expert on the Go compiler internals but I could probably get up to speed pretty quick. I certainly know the language very well.
I do believe we had a working port once as well but would need to explore the history on that.
Obviously if this sounds good I need to register as a mentor so @PulkoMandy please let me know the details on that, either here or by email or PM.