I honestly did not expect that my review would have such an effect on the system development. In other words, Josh (the person behind the nickname Jggimi) decided to change the whole approach to his system based on my critique. It was one of the most constructive reactions to my reviews I’ve seen so far.
Let me take a little break now, and let Josh explain his future plans. I quote his e-mails, slightly edited.
First, I'm not a developer. I'm just an OpenBSD user who shares preconfigured "live" images as a service.
Not only am I not a developer, these are not OpenBSD -- they are .iso images which are OpenBSD-based. It may be a subtle distinction, but OpenBSD is not distributed in these forms. And there are small modifications to the system -- changes to the rc(8) script -- which mean that it is not exactly the same as OpenBSD.
I began writing a letter trying to address individual points in your review... but also began rethinking what I should produce as live media.
The 5.1-release source code has been tagged, and will be readied for CDs, download, and support on or about 1 May. For my part, I have begun building 5.1 binaries and thinking. Thinking a lot.
When I initially created live media, I built it to include various X Windows managers, eventually settling on the built-in fvwm, fluxbox, XFCE, KDE, and Gnome. For each of these, I used the default configurations, ensuring that a terminal and a browser could be easily reached. Nothing more.
These were never intended for a production use -- none of them. They were for system rescue (from OpenBSD users familiar with configuration tools), for hardware testing -- typically to see if a new computer's hardware could be used without having to install -- again, by an OpenBSD user who might be in a store testing some equipment before purchase. Lastly, for users unfamiliar with OpenBSD who might want to see what it was like. To that end, I added the popular window management environments XFCE, Gnome, and KDE. I wanted them to know that the GUIs they might be used to from another *nix would work for them on OpenBSD. These DEs grew and grew over the years, and were less and less useful through the performance bottleneck of being iso9660 filesystems.
Unfortunately, I made two errors of omission -- both of which you uncovered for me. The first was scripted WiFi support, which I don't have; the second was not ensuring I had crafted well functioning GUI environments without cul de sacs, dead ends, or missing tools. The latter was not very important to me, as these were not intended for production. Your tests, though, indicated I should have eliminated errors and limitations in configuration.
It is very easy to install OpenBSD onto USB devices, so I never bothered to create "liveUSB" media images. However, there is a need for those who travel across borders with electronics to keep private information network-attached, not on their devices. So now I am considering "read only" USB images, for production use.
No decision has been made, yet. I'm still considering what the impact of this will be -- primarily taking a non-production, test tool into a production use realm.
But I wanted you to know that your review was the impetus for this. I'll let you know what I decide, once a decision has been made.
My build process has begun -- I am building 32-bit and 64-bit versions based on 5.1-release and the two thousand or so third party packages that would normally be used to configure the dozen .iso images.
I am not yet sure which configurations I will keep, or discard -- if I decide to make appreciable changes, or change the purpose. If I want to publish these in sync with the OpenBSD project I'll need to make my decisions in time to configure, test, and publish by 1 May.
What can I add? Not much actually. Except that I am absolutely flattered by the impact I made on that system.
Let’s wait for couple of months. If everything goes according to the plan, I should come back to Jggimi’s Live version of OpenBSD Live somewhere in May.
In the meantime… stay tuned!
0 comments:
Post a Comment