Wednesday, September 28, 2016

Why alpha 2 isn't launching today

The project blog has been quiet for a while, as I've continued to focus on alpha 2. Poem is my first software project of this size, and it is challenging to be crafting it solo. Because of this, Poem has been teaching me valuable lessons and has been quite the crazy adventure for me since I started it... as reading the project's history on the about page will show for itself!

But, here we are -- right now. The next release was scheduled to launch as Alpha 2 before the end of this month into mid-October... but to be honest, here are the two major reasons we're just not ready for Alpha 2 to launch today... or 1-2 weeks from now.

1. Desktop work

Right now, Couplet apps still do not have a menubar, there are no per-app shortcuts, no usable desktop, and other missing details that I consider solid requirements for alpha 2 to be ready to launch.

But the other reason that alpha 2 for Couplet had pinned me down is that I was trying to stuff the whole environment into Firefox. It sounded awesome when I thought of it, but window movement and usage was limited to one canvas, which meant that windows were crammed in one area, amongst other issues I saw with that model working on a full desktop. Recently, I've decided to alter course, and am going back to using paired with a more compact browser as a backend for powering the new Couplet apps instead, as building it this way has proven to be superior. At the moment, though, it needs work.

And as for Soliloquy (the Haiku distribution that powers Vintage), it would still ship based on Haiku alpha 4.1 gcc2 if it did launch today. Whether or not we will switch to a more recent Haiku revision as the base for the image is still something that I'm actively debating before releasing its next iteration to the public. While I appreciate the long work and the efforts the Haiku project has done (and still is doing) with the Depot, I also am not willing to follow the package centric ideal in Poem, as I believe it takes away freedom from the end user to control the contents of his or her filesystem -- something that BeOS and earlier revisions of Haiku believed in... and I still believe in.

2. Mobile work

As a sidenote, Poem really is aiming for a mobile edition of Couplet that will run on Pi with the next release, but it's also not ready for this stage either. To explain this, while you could run Couplet on the Pi on a full-size monitor or TV as a desktop-like experience, elements do not scale down to provide a good mobile experience. On the average 5-7" display with a 640x480 or 800x600 resolution most kits include for these computers, realistically, they need their own UI. This is why (at least in my opinion) having the everyday Linux desktop is a royal pain on these the same sense running CE 2.0 on H/PC was. To realistically solve the small screen problem, Poem will need to have a separate mobile shell that will run applications fullscreen and without needing to have a traditional desktop. And so, in a sense, Poem is essentially 3 big subprojects (Couplet, Soliloquy, mobile) with one goal because of this realization. And even if we decided to ignore any plans toward mobile for now and upload and launch alpha 2 anyway today, it would still not offer the experience on the desktop I'd expect.

No alpha 2

So, there it is. I apologize for deciding to halt this release; I know that there's been anticipation over it, and people reading the project's blog will probably be disappointed after waiting since the project's rebirth in May for a fresh build to have fun playing with.

But please know that I really am dedicated to finishing a system that has the look and feel of the Be desktop, as I believe that both the BeOS and MacOS were the simplest (and best) desktop systems ever built. While I definitely do not want Poem to slip into vaporware lore or to have no plan to it, I also want it to be a passionate effort that doesn't crank out junk.

That said, I just wanted to be open with everyone and explain what's happening with Poem. The next release really is on the way... it's just going to need to wait longer than I had hoped for. Right now, my plan for Poem is that we will get a new release out once I can conquer at least a few more areas I see hindering it from launching... and I'm not sure exactly when this will be done, but I am hoping to try again for releasing a new build before November and possibly move up to beta if enough features are finished... but only time will tell.

Thank you for reading -- and for your patience!

-- Andrew Greimann

Tuesday, August 9, 2016

Roadmap for alpha 2

Now that Poem is back online, work on Alpha 2 has quietly been resuming from the point where the project was almost quit for good. Alpha 2 is going to be a huge milestone for the project, and a significant leap forward from Alpha 1 once it is released. Current plans for the upcoming preview (a2) builds are:

1. "Aubade"

All major Poem projects and editions are named after poetry; an aubade is an ode or a song to the dawn. Aubade is the codename for the all-new UI which will run utilizing Firefox as a backend. It will power all new Poem (Couplet) applications. The UI be heavily inspired by and modeled after Dano -- but will also take tablet computers into consideration as well in its design as a hybrid.

Please do not expect it to be complete or do much in its first appearance in alpha 2, but it should prove to be an exciting starting point for all future builds once it debuts.

2. Deskette

One of the primary goals for Poem is to create a new environment inspired by Be, but the problem has been that up until now, Poem hasn't been able to truly start doing this yet. Starting with alpha 2, early builds of Deskette and Pointer will be introduced to Couplet to imitate and be elementary substitutes for the Deskbar and Tracker.

3. Vintage

The previous 'Throwback' concept from alpha 1 has been merged into a new internal project named 'Vintage', and will include several major improvements as an active frontend to qemu for running Soliloquy in a virtual machine. With the upcoming preview, Vintage will be able to set various preferences. For example, you'll be able to define the memory size of the guest, the number of cores you'll be using, and whether to use acceleration. You will also be able to create and delete virtual hard disks.

4. Finding a new core

And finally, this was a really huge decision, but one that had to be made before going any further.

After evaluating the performance of alpha 1, it was clear that Poem needed to find what base it would use for R1. Debian Gnu/Linux simply could not be what defined the main edition of the Poem project, mainly due to increasing support around systemd, along with other reasons.

Several systems for the new base for Couplet were considered, including illumos, NetBSD, and several distributions of Linux. illumos is an excellent system that almost became my final choice, as it has an awesome feature set and legacy. In the end, however, I found flavors of it didn't work well with Vintage or the x86 tablets I tested with it. While NetBSD was also a good candidate I considered due to its incredible manageability and ports, several new machines didn't play that well with the release I tested (i.e. my MacBook and HP Stream Pro). And so, after reconsidering whether to resume building Couplet as a Linux-based system, I decided to evaluate a small handful of distributions that I considered to be qualifiable candidates.

Finally, after long hours of testing and comparing potential systems to utilize as a base, starting with Alpha 2, the Couplet edition will transition to Void in order to have a much more solid release that meets the goals of the project for R1. Like Slackware, Gentoo, and a select few others, I find that though Void is small, it is systemd-free, well-built, extensible, very configurable, solid, and so far, has been performing well.

That said, while Poem cannot (due to project constraints) pursue more than two base systems (Haiku and Gnu/Linux)... if you really don't wish to use the Linux kernel (or just miss Debian), a kFreeBSD i686 flavor of Couplet will also be maintained as well for those that want to escape the norm.

5. Release estimate

Alpha 2 is currently planned to be released as a live image for both editions sometime this Fall, between early September to mid October 2016, depending on the speed of development. Please keep following the Poem blog for any new updates or changes!

So there it is, and I'm really excited about what's coming next!

-- Andrew Greimann

Monday, July 18, 2016

Hello again, Poem

This post was actually supposed to be a farewell. I had completely quit Poem back in March, after becoming discouraged with it for varying reasons. But now that the domain is within days of expiring, there is a part of me that isn't willing to just let it fade into oblivion.

I kept thinking of how I could apply the ideas for 1.0 into a whole new project instead of this one, but each time, my mind jumped back to Poem. I then realized that Poem was meant to be just one thing: something I'd love looking at every morning. It didn't have to be anything else. This is the critical part I missed before, and if any project (like this one) doesn't have that one idea behind it, it's not worth asking people to even read about it.

That said, Poem will not endeavor to be another Senryu, Haiku, etc. If you are hoping for that sort of thing here, kindly use Haiku. While Poem will continue to use a Soliloquy image (distribution) for the classic VM, its primary goal will remain different - to create a new UI modeled after Be on a new foundation. The accompanying mission of Poem will be to focus on what's next, and if that means rethinking how something works, tinkering with new ideas, and choosing to go with them, then that is what Poem will do.

I say all this because as of right now, I've decided to take back the helm of the project I left... and I'm truly hoping you'll love what's ahead. :-)

Thank you, all.