nanog said: Awalton, I did not intend to denigrate the fine work you and others are doing to make Gnome/Ubuntu better. I already file bugs and you have motivated me to make another financial contribution to the Gnome Foundation (http://www.gnome.org/friends/). I also specifically requested that the money be targeted towards development of GvFS. Nevertheless, it is my opinion that GvFS is not ready for a production environment or an LTS release. I have absolutely no problem with it being included in Ibex.
You can throw some money my way too ;). But back to the point: GnomeVFS is considered EOL and deprecated. Not shipping it would mean Ubuntu would be the sole supporter of it for the next.. 5 years. GnomeVFS had two active developers before GVFS was invented. GVFS now has 6 that I can count on the tips of my fingers, and we've had patches from probably 20 contributors. Quite frankly, if I had to support something for the time Ubuntu has to, I'd choose GVFS every time.
nanog said:
I am most concerned about regressions from Gnome VFS and Gutsy/Dapper.
As are we, and they're being worked on, on a case-by-case basis. Filing bugs and marking them as regressions is a very quick way to get the upstream to notice. We know that we lost a significant amount of functionality; we went from an old, mostly finished but entirely buggy, to the new, partially finished but a lot more stable system. I've already mentioned this in one other thread, so I better knock on wood before I say it again: We have had no crashes reported to the Upstream GNOME bug tracker of GVFS or its related components since its release. We still to this day get about 10 crashes of GnomeVFS a day, and this is after years and years of work.
nanog said:
*Launchers, guis, network shares that still disappear on upgrade.
Launchers shouldn't disappear, that's not a GVFS issue. GUIs shouldn't disappear either, and again, not a GVFS issue (though it should be noted that there have been changes in the GUI to better support GIO/GVFS features vs. those available under GnomeVFS; changes aren't always regressions).
Network shares disappear because they're completely incompatible with GVFS. GnomeVFS is stateless and shares never were mounted, and were tracked through a bunch of pseudomagical text files. GVFS shares are stateful, and a tracked my mounted locations in a folder, as well as my the mount daemon gvfsd, and by the individual protocol handler's mount daemon (gvfsd-network, gvfsd-smb, etc). I'm not really sure what a "migration UI" would even look like in this case; as long as you kept your server information, it shouldn't be a problem to reenter the information and have a new Bookmark for it.
nanog said:
*Functional smb shares (this has gotten alot of negative attention in the press).
Works for some, not for others. Some aspects of SMB just aren't handled yet. Others are just plain bugs which will be worked out over time. And they may not even be our bugs in all of the cases, we've had some reported issues with the smbclient library we depend on as well. Unfortunately, the person who wrote and understands most of that code is on leave, I don't know anything about SMB and it's a dontcare for me anyways (I'd rather not use Microsoft created protocols for something that can be done much simpler anyways).
nanog said:
*Stable SSH, Webdav, and FTP.
We've had only one or two issues with SSH, which we're tracking. Very odd that some servers don't work... WebDAV isn't done and this is a known regression. And we're using FTP code that we share with Mozilla so essentially, we're either blocking on them fixing it, or on us fixing it together. There's some really great FTP code out there that we just can't use because GVFS is LGPL and not GPL though, and there are some thoughts that we (well, I should say "I", because I'm the person considering this at the moment, though it has been discussed with the author of the FTP backend Benjamin Otte as well) might pull a David Zeuthen and link to it anyways (for understanding of what I mean by this, click here).
nanog said:
*Fuse mounts that do not crash spontaneously and upon logout. *Permanent mounting of dead network connections. *Fuse mounts shown as root in nautilus.
You have three whole bugs there that we aren't tracking. If users don't report them, we don't know about them.
nanog said:
I sincerely hope that someone is paying close attention to what other distros are doing.
Whatever person wants to do this is fine with that person. But Ubuntu, the organization, is not chartered to follow the lead of others with the important caveat of Debian, but that's more out of respect and cooperation and sharing of labor, and in a lot of ways we lead them. Like said, Ubuntu doesn't have the staff to have an "Official Inter-Distribution Communications Council/Liaison" or anything of the like, which is my point on this. If you'd like to do this, please join Ubuntu and start it, and get some help, you'll need it.
Ubuntu talks to upstream in their various components, and various people upstream share information, namely because the upstream components are what's shared between distributions. But that's about the extent of this communication in practice. This is so much to the point, that downstream deltas on pieces of software can get drastic (see the news about the Debian/Ubuntu SSL bug). Point be told, this was what Launchpad was partially invented to solve, but unfortunately the distro-barrier still exists.
It's also factual that there is no competition; Ubuntu doesn't have the energy to say "Haha, we're ahead of everyone else," or to fight to be there. The idea is not to fight with other distributions, because that's a whole lot of wasted effort on everyone's behalf. Ubuntu syncs with what's available, this is a fact, and it's been reiterated by Shuttleworth just today. The fact that Ubuntu has to ship betas every now and again is not because they're trying their hardest to be bleeding edge, it's because that's where the support is.
Firefox 2 will receive very little of the developer's energy, so both Ubuntu and Fedora chose to ship Firefox 3 early. In a year, you can expect there to be virtually no support upstream.
Likewise, GnomeVFS will receive no energy, so all three major GNOME distributions chose to ship GVFS (Debian did not, and we receive a steady flow of duplicated bugs from their users). ESound has been dead for years, so Ubuntu chose to ship PulseAudio. Quite frankly, there wasn't a lot of choice in the issue; if you want audio features like Windows and Mac OS X have, you will ship PulseAudio (and it's been a blueprint to clean up the audio situation in Ubuntu for years now; this is a huge step towards that cleanup).
nanog said:
It is amusing that the wikipedia link you provided mentions that "call and response" may have originated from scottish-irish protestants (at least in the usa). Because if it's on Wikipedia, we know it to be a fact, right? Call and Response has been practices by African people for hundreds of years now, and is deeply ingrained in African culture and in African music (which, interestingly enough, is how Rock and Blues now has it in America), and that's is all that is meant by it. Free software emulates this Call and Response, as the developers Call, and the users Respond. And I thought it fit the drum metaphor quite well. (That, and I'm of partial African descent and it's one of the parts of my heritage I love to learn about.)