Niches for the Hurd

In the bug-hud mailinglist we did a search for niches where the Hurd is the biggest fish in the pond.

This search was segmented into four distinct phases, three of them major:

  • Brainstorm
  • Reality check: can already do vs. could be used for
  • Turn ideas into applications
  • Find a compromise -> About which niches should we talk in the wiki?

Brainstorm

"Which niches could there be for the Hurd?"

Basic Results

The result is a mix of target groups, nice features and options of the Hurd, reasons for running a Hurd and areas where the Hurd offers advantages:

Nice features and options the Hurd offers

  • Give back power to users: arbitrary mounts, subhurds
  • Nice features: dpkg -iO ftp://foo/bar/*.deb
  • Easier access to low-level functions
  • Advanced lightweight virtualization
  • operating system study purposes as its done with minix
  • The possibility to create more efficient and powerful desktop environments
  • Having a complete GNU System
  • All-in-one out-of-the-box distro running a webserver for crash-proof operation.

Target groups and strong environments

  • Tinkerers who like its design.
  • multicore-systems

The keyphrases in more detail or with additional ideas

Give back power to users: arbitrary mounts, subhurds

Simpler virtual computing environments - no need to setup XEN, everyone can just open up his/her computer for someone else by creating a new user account, and the other one can login and easily adapt the system for his/her own needs. If most systems just differ by the translators setup on them, people could even transfer their whole environment from one computer to another one without needing root access or more root interaction than creating a new user account. "I want my tools" -> "no problem, just setup your translators".

Also it would be possible to just open an account for stuff like joining the "World Community Grid" allowing for easier sharing of CPU time.

Easier access to low-level functions

"One important use is for very technical people, who don't always go with standard solutions, but rather use new approaches to best solve their problems, and will often find traditional kernels too limiting."

"Another interesting aspect is application development: With the easily customized/extended system functionality, and the ability to contain such customizations in subenvironments, I believe that Hurd offers a good platform for much more efficient development of complex applications. Application developers can just introduce the desired mechanisms on a very low level, instead of building around existing abstractions. The extensible filesystem in particular seems extremely helpful as a powerful, intuitive and transparent communication mechanism, which allows creating truly modular applications."

Advanced lightweight virtualization

"There is also the whole area I called "advanced lightweight virtualization" (see http://tri-ceps.blogspot.com/2007/10/advanced-lightweight-virtualization.html ), i.e. the ability to create various kinds of interesting subenvironments. Many use cases are covered by much bigger fish; but the flexibility we offer here could still be interesting: I think the middle grounds we cover between directly running applications, and full isolation through containers or VMs, are quite unique. This could simplify management of demanding applications for example, by partially isolating them from other applications and the main system, and thus reducing incompatibilities. Creating lightweight software appliances sounds like an interesting option."

The possibility to create more efficient and powerful desktop environments

"While I believe this can be applied to any kind of applications, I'm personally most interested in more efficient and powerful desktop environments -- these considerations are in fact what got me seriously interested in the Hurd.

Even more specifically, I've done most considerations (though by far not all) on modular web browsing environments. Those interested can read up some of my thoughts on this:

http://sourceforge.net/mailarchive/message.php?msg_name=20080909073154.GB821%40alien.local

(Just skip the text mode browsing stuff -- the relevant part is the long monologue at the end... I really should put these ideas into my blog.)"

Nice features

Another example of features which would be easily possible with the Hurd:

  • transparent ftp (already possible!):

    • settrans -c ftp: /hurd/hostmux /hurd/ftpfs /
    • ls ftp://ftp.gnu.org/
    • # -> list the files on the FTP server.
  • media-player translator:

    • settrans play /hurd/mediaplayer_play
    • cp song1.ogg song2.ogg play
    • # -> files get buffered and played.

or even:

  • cp ftp://foo/bar/ogg play

that's KDEs fabled network transparency on the filesystem / shell level (where it belongs to be desktop agnostic).

  • add temporary filesystems anywhere via settrans -a NODE /hurd/ext2fs

  • On-demand mounted filesystems via a passive translator which unmounts the filesystem when it isn’t used for some time.

  • make everything temporarily writeable without really changing it via unionfs. Store the changes on an external device.

  • Read tar archives and mbox files via ls foo.tar.gz,,tarfs and ls foo.mbox,,mboxfs, respectively → nsmux.

  • Use stuff like the new akonady (personal information) framework in KDE more efficiently from the shell.

Reality check

Check which of the ideas can already be done easily with the Hurd in its current state, which ones are a bit more complex but already possible, which ones need a bit of coding (could be accomplished in a few months judging from the current speed of development), which ones need a lot of work (or fundamental changes) and which ones aren't possible.

Already possible and easy

  • Sample translators:

    • hello world.
    • transparently bind FTP into the filesystem
    • hostmux + ftpfs -> connect to FTP automatically via asking for a dir named after the hostname -> fully transparent FTP filesystem: "touch ftp: ; settrans ftp: /hurd/hostmux /hurd/ftpfs / "
    • bind any filesystem at any place in the directory tree (you have access to) without needing to be root.
    • elegantly mount iso images and similar as unprivileged user.
  • Other useful stuff:

    • Install deb-packages from an ftp server via 'dpkg -iO ftp://foo/bar/package.deb'
    • remount a filesystem readonly as regular user: fsysopts /foo -r
    • give a process additional group and user permissions at runtime:
      $ groups
      root
      $ ps -L # gives me the PID of my login bash -> bashPID
      ...
      $ addauth -p bashPID -g mail
      $ groups
      root mail
  • Having a complete GNU System (but not yet on every hardware, and only about half the software Debian offers has been ported).

Already possible but complex or underdocumented

  • Easier access to low-level functions via translators.

  • Operating system study purposes as it's done with minix.

  • Tinkering for fun - need documentation about the fun things which can be done.

Need a few months of coding

  • A filesystem-based package manager.

  • subhurds for regular users

    • A framework for confining individual applications is really just one possible use case of the hurdish subenvironments. Writing the tools necessary for that should be quite doable in a few months. It's probably not really much coding -- most of the work would be figuring out how it should be set up exactly.
    • subusers
    • "subdo":
      # Example: Let a virus run free, but any effect vanishes
      # once the subhurd closes.
      $ subdo --no-lasting-changes ./virus
  • subhurds for quickly adapting the whole system without bothering others.

  • Define your personal environment via translators, so you can easily take it with you (translators written in scripting laguages can make this easier - they could also for example be taken to each computer on USB stick).

  • A more powerful alternative to FUSE filesystems: While FUSE is limited to standard filesystem semantics, while Hurd translators can implement whatever they want. It is possible to change the behaviour in any aspect, including the way file name lookup works. Admittedly the only specific use case I know is the possibility to implement namespace-based translator selection with a set of normal translators, without any changes to the Hurd itself. It is also possible to extend the filesystem interfaces, adding new RPCs and options as needed. This allows using the filesystem for communication, yet implementing domain-specific interfaces where standard filesystems are too unefficient or cumbersome. A sound server would be one possible use case.

  • Namespace based translator selection (if you for example want to quickly check the contents of an iso image, just look at them via 'ls image.iso,,iso9660fs').

Need a lot of coding or fundamental changes

  • Effective resource management (For example via Viengoos on which Neal Walfield is working). The idea is that we could make a virtue out of necessity: Once we have a proper resource management framework, we should be able not only to catch up with traditional systems in this reagard, but in fact surpass them.

  • The possibility to create more efficient and powerful desktop environments.

  • Currently to offer CPU time to some project (like the World Community Grid), it is necessary to install a program from them, and they can then do only what that proram allows them to - which leads to reinventing a processing environment instead of just using the existing OS. With the Hurd people could just create a user for them, give that user specific permissions (like "you're always lowest priority"), add the public ssh keys of the project they want to donate CPU cycles to, and the project could just turn the computer into the environment it needs for the specific computation, without compromising the main system in any way (needs better resource management).

  • A shared MMORPG game world consisting simply of files for levels and person descriptions with access rights. All synchronizing is done on the translator level. Programs only have to display the given files and quickly update the state of their own files, so the programs stay very easy. The translator could notify the program when something changes.

  • Multicore systems (need to fixup Mach for SMP)

  • Running parts of the Hurd on different computers, maybe even with shared servers on dedicated hardware (Cloud Computing when the servers can be made to migrate from between computers). Maybe this should be placed in "need a lot of coding".

Unfeasible ideas

Applications

A minor phase, which will surely be interleaved with the others: Making the ideas tangible to turn them into ways how people can use the Hurd.

"Hey, look, this is the Hurd. You can use it like this to do that which you can't do as well/easily/elegantly in any other way."

Applications for private use

Applications for companies

How an application should be presented so people can easily test and digest it

We need stuff which gets people to say "hey that's cool!"

And it must be readily available. If I have to search for arcane command line parameters before I can use it, it's too hard.

From what I see, each direct cool application must be about as simple as

$ qemu hurd-is-cool.img
$ login root
$ settrans cool /hurd/cool
$ ls cool

One main focus in this example is: No command line parameters but the ones we really need. No "-a", if the example is also cool without it. No "--console" if it works otherwise.

Especially no "qemu --cd livecd --hda hurd.img ..." - that one is great for people who already know qemu or want to learn it, but the goal here isn't to teach people better usage of qemu, but to show them that the Hurd is cool, and only that.

All that interesting advanced stuff just gets newcomers confused.

The translator concept in itself is enough news to faze a mind - anything else can easily be too much.

If the application isn't as simple as the example above, then the best step would be to see if we can make it as simple - if that involves writing trivial scripts than be it so. They are trivial only to those who already understand the underlying concepts.

And now enough with rambling :)

The Hurd is cool, and the complex to use applications are cool, too. But they are hard to present in a way newcomers easily understand.

Compromise

For each niche:

  • What do we have to do to conquer the niche?
  • How many additional programmers can the Hurd get in this niche?
  • How does choosing this niche limit the flexibility of further development (for example due to the goals of the people who join up)?
  • Can we easily move on to conquering the next niche once we got this one?
  • What should the Hurd accomplish on the long term (long term goals)? Which possible niches help that?

Each participant:

  • Give your personal priorities to the niches:
    • Must -> all of these of all developers must be included; remember that at most 3 to 4 ideas can be conveyed in any text.
    • Should -> The number of shoulds can be used for ranking and similar.

("must", because in a community people can do what they perceive as important, and telling someone to stop what he's doing is no option (in my opinion))

Result: We talk about the niches we can already fulfill :)

Things to do

todo-item -> niches for which it is useful.

This might be useful for the next GSoC.

Easy

  • Port debian packages to the Hurd -> currently mainly tinkerers, but also any other niche. In the long run this is necessary for every user. Easy start for devs.
  • Document easier access to low-level functions via translators, one function at a time. -> tinkerers.
  • get nsmux ready for regular users by setting it up in the LiveCDs by default. -> show tinkerers what it can do.

Complex

  • A filesystem-based package manager: Unionmounting packages. With filterfs from nsmux packages any user should be able to selectively disable any package without affecting the system of others. Simple active translators can add packages. -> clean design and more freedom for tinkerers to setup test environments: “Does this also work with XY disabled?”
  • Enable subhurds for regular users via a subdo command: A framework for confining individual applications. -> tinkerers for testing their work.
  • Define your personal environment via translators, so you can easily take it with you ⇒ system on a USB stick. Would work great with a filesystem based package manager. -> ?

Huge

  • Get Hurd/GNU Mach ready for efficient multicore usage. -> multicore
  • Running parts of the Hurd on different computers, maybe even with shared servers on dedicated hardware (Cloud Computing when the servers can be made to migrate from between computers). -> multicore on steroids :)