On Tue, Sep 16, 2008 at 10:33 PM, <[EMAIL PROTECTED]> wrote: > Just adding my $.02, since I'm actually the one doing most of the > transport work. Shawn has made most of the relevant points already, but > I'm going to expound upon a few. > >> > The correct approach would be to eliminate the repository/depot and >> > just make packages be files that are put someplace. Whether that be a >> > web server, ftp server, nfs server, CD, usb stick, or local disk is >> > immaterial. > > I think you're confusing the correct approach with the approach you find > simplest to comprehend. I haven't read any articulate objection that > indicates we made an incorrect choice when favoring network install. >
The network is not ubiquitous as it may seem. Rather than declare the IPS team to be incorrect, I'll simply say that believe it or not, there is a large user base which does not have network access, and if future opensolaris releases are going to become network centric, then these large user bases will get cut off. We're simply stating our concern that network-only installs would affect opensolaris' reach later on. However, your messages later in this particular post lead me to believe that a depot-server independent may indeed be possible. There are doubtless many FUD thoughts about IPS. Perhaps all that's needed is inputs from us users and answers from you developers :) >> Packages are just essentially files that are put somewhere right now. >> It's just that the all the individual files of the package are put >> somewhere instead of a single file. Once package dependencies and >> grouping has been redone properly, users will benefit from reduced >> bandwidth usage and I/O. > > Users already derive some benefit. When we perform an upgrade, you only > download the files that have changed. Other packaging formats require > that you download an entire package with all of the files, changed or > not. > Ok, this is surely nice to know ! :) > > We will have these installation sources, they're a degenerate case of > the network installation. Once we re-write the transport layer to use > HTTP 1.1 and pipelining, we'll have all the pieces that we need. The > libraries that we will use, instead of Python's underwhelming urllib, have > support for HTTP, HTTPS, FTP, FILE, and on and on. > And such developments would be really exciting to see ! :) >> > The current mechanism is a huge barrier to entry. Not only does it >> > make it much harder for other sites to distribute your software, it >> > makes it enormously hard for end developers who wish to distribute >> > software for OpenSolaris. If, and this will be the common case, they >> > aren't able to run their own repository then they can't create or >> > distribute OpenSolaris packages at all, which harms the whole >> > ecosystem. > > We aren't stopping anyone from running their own repository. Here's what Moinak means. Unless the IPS client can pick files off http/ftp/file, community distros who want to use IPS will not be able to request various public sites to mirror our packages. This is because public mirrors only run http/ftp/rsync. And file:// would be absolutely awesome for someone who gets home the latest set of packages via a CD-R from a friends house or a User Group meet. Your last statement in this post is reassuring on this front, I tell you. > >> > This is why the on-disk package file format is the crucial piece of >> > the puzzle, because that should be the form in which the software >> > moves around. >> >> I don't know why you insist that can be the only form that software >> moves around in -- it's certainly a valid form, but not the only form in >> my view. I can tell as a sysadmin what Moinak means by this - For worst case scenarios, I need to be able to pick up installables and install them in case my depot server or http server is not available for some reason. I've mentioned a case on another response in this thread, where I once had to grab gcc and ruby off my repository file system and install them - I couldn't sync from my upstream servers in my other offices. > > I agree with Shawn. There's no reason why we need to insist upon only > one way in which the bits get moved around. > >> To quote j: >> "The download format is going to be changed as part of other transport >> work. Our intent is to move away from tarfiles, and instead have >> clients perform pipelined HTTP GETs." > > And once we get to that point, we should be able to remove what little > logic remains in the depot for downloading files. The goal is to have > the download front end be interchangable. You could use Apache, > CherryPY, your HTTPd of choice, or even FTP or file. > Believe me, this was the most reassuring set of statements that I have read in this entire thread :) As an end user, my feedback would be that this depot-free mechanism that J's talking about will become very popular and help make IPS more acceptable (assuming that other issues are also ironed out). -- Sriram _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
