RFC - Centralised OXP hosting

[DaddyHoggy]

Umm it might be a bit late to ask this, but might it be worth making the distinction between central hosting of OXPs, and central databasing of OXPs?

For the most part people use box.net or equivalent for hosting OXPs where they don't have their own sites, and generally that works well enough. It only goes wrong when people let their accounts die, which is usually a sign that they're not into supporting OXPs any more anyway.

In my opinion what is lacking is not the hosting per-se, but the way of searching/updating OXPs, which has nothing to do with hosting them. The weak link at the moment is the OXP list on the wiki, as it's difficult to control and update and a pain to use for checking you're up to date with everything.

Plus also on a personal note would it not be better to ask OXP makers if they actually want to host their stuff elsewhere? Personally I like having my stuff still under my control, plus all of the stats etc that box.net provides. I'd have to hear a compelling argument to move them elsewhere, as at the moment for me box.net isn't broken so I don't see the need to fix it by hosting stuff elsewhere...

Unless you're paying for the Pro account - aren't you limited to 10MB file size and limited bandwidth on box.net? For some OXPs this could become a problem - although I completely understand the wish to maintain absolute control of your own work.

[ADCK]

Plus you could get that same "absolute control" on this theorhetical central site anyway, without a 25mb limit, and the ability to add comments/decriptions/screenshots/etc which can't be done on box.net

[Cmd. Cheyd]

Unless you're paying for the Pro account - aren't you limited to 10MB file size and limited bandwidth on box.net? For some OXPs this could become a problem - although I completely understand the wish to maintain absolute control of your own work.

This is the situation I'm in with SR2. When it's released, it'll contain 70+ textures, and have 2 optional download "texture packs" (one Hi-Res, one Lo-Res). Both the standard and the Hi-Res versions will be WELL beyond the free Box.Net limits. Hell, with 70+, the low-res may as well. For testing and early sharing with some select folks, I've set up a home FTP server. But that won't hold up under heavy downloads.

I really like the stats and controls available with Box.net. I'd LOVE to have the functionality of the central repository. Is there any chance we can get the best of both? The repository's centrality and customization (and hopefully automation, please please please?), and Box.net's stats / feedback?

[davcefai]

This "absolute control" thing has me puzzled.

As far as I am aware OXPs are published with a Creative Commons licence. Once the first copy is downloaded, that's it. It's out in the wild and can be copied and further distributed under the licence conditions.

The only control left to the author is to remove the OXP from wherever it is hosted but anybody can host it somewhere else.

[Thargoid]

Absolute control - That wasn't quite what I meant. It was more being able to keep track of your own OXPs downloads, being able to temporarily withdraw them if problems are found with them (or permanently if required), and generally to know what's going on. And as for licenses, without wishing to open all that debate again, not all OXPs are licensed the same...

Abandoned OXPs - The point is certainly true, but we're then in the scenario of when do OXPs become abandoned? We've had a few examples I can recall where OXPs have been considered abandoned then their authors turned up again.

OXP size - Yes there are limits to what box.net can do, but I never said box.net should be the only thing used to host OXPs. Plus not everyone has huge download allowances, broadband or disc space/fast machines to run large OXPs. So it might even act to keep things in sensible file sizes.

Oh and am I the only one who gets nervous about a plan to put all OXPs "in one basket", in case said site goes down or gets Oosat2'd?

Like I said before, I'm not against the idea totally, but for me the current hosting solution I have isn't broken so I don't feel any need to fix it. What is broken is the wiki OXP listing solution, and for me that's where efforts should be focussed. Until I can be convinced that a new solution would be better than the box.net method I'm currently perfectly happy with, then my stuff is staying exactly where it is. Sorry but being an engineer I tend to work to the old adage about things that aren't broken.

[ADCK]

For me it's about professionalism, crappy wiki + 3rd party decentralised hosting = very amatuerish.

I'd take a centralised oxp hosting with all the benifits of box.net +more over a what we have now... which is basically nothing. A small subsection of an unmoderated wiki and 3rd party hosting.

Not broken? Can't break what doesn't exist

Fixing the wiki would only be a band-aid solution, and since it's not even OUR wiki what's the point?

People don't seem to grasp that the theorhetical oxp hosting site could be so very much more better than what we have now, no clutter, no download/upload size limitations, ability to make comments/screenshots, search functionality, top 100 oxp's, moderators, having an actual say in development of the site, not having to mooch off of alioth.net and box.net, + many more, are just some examples off the top of my head.

I understand that some of you are in a nice comfy little rut and don't like change, but that's not progression, it's stagnation.

[Selezen]

I have 5GB of webspace currently sitting doing very little. I also have a wiki site set up on it that anyone can register on and use. It also has unlimited bandwidth.

As I've said all along, people are welcome to use it for file hosting.

[Cmdr James]

The real question here, with respect, is what happens when the person running it gets bored (becomes a hermit, has children, gets run over by a bus, etc.)?

A nicely managed single point of reference is nice (with some caveats), but only for as long as it exists and is managed. There was nothing fundamentally wrong with oosat or oosat2. There is also, in principal nothing wrong with the wiki for this.

It is important to remember that there may be a "common" or "default" area for OXPs but it cannot, by definition, be exhaustive. There will always to works in progress (some never finished), private modifications, various version (with different textures perhaps) which may not all be public, some shared by email, and some licensed in a way that prevents centralising.

As with most things in life, this isnt fundamentally a technology challenge, it is an organisational one. Get a few people committed to tracking and keeping $resource up to date and it works, adding drupal on ooxp.oorg may make this easier, but it will not change the fundamentals.

[bigmike20vt]

re the larger file sized oxps.....

could some sort of peer to peer file sharing network be put in place?

i have a pc which is on on average maybe 6 hours a day which already has utorrent installed, i certainly would be happy to put aside a gb to share a torrent with a bunch of larger oxp's in.

so long as enough people were happy to do this, it would in theory keep the file alive.

i realise this has its own caveats with not up-to-date files but if a date was menioned in the torrent name then it should be simple to keep on top of it.

just a thought......

[Diziet Sma]

Wow.. lots of good ideas and opinions showing up now.. this is what I wanted to see.

Ok, it's time for me to elaborate on my vision and deal with the points raised above.

Control:
OXP authors will each have their own account. They will have full editorial control over the page/s dealing with their OXP/s, including content/images and the ability to upload/delete their own files as they see fit.

If an author really does not want to have their work hosted on the central server, they can instead simply have their page/s link to their own hosting server. This way they will still be able to take advantage of the centralised OXP database and automatic notifications (see below).

Statistics:
I plan to provide the same or better information that you have with box.net or whomever you are currently hosting with. If there is specific additional info you'd like to have, please let me know.

Abandoned OXPs:
I would like the repository to serve as a method by which these can be saved from fading into oblivion. Should the original author re-appear, control will of course be turned over to them, if they so choose.

Crashes/corrupted data:
1and1, my hosting company, make a backup of my sites every day. Being a belt-and-braces kind of guy when it comes to important data, I also intend to set up my own regular backups (database and files) on off-site storage. (apart from anything else, this makes it easy to change to another hosting company, if it becomes necessary)

Automatic mass downloader/updater for OXPs:
Something like this ought to be quite feasible with a centralised repository.

File sizes:
A number of OXPs are now exceeding the 10/25MB limits imposed by free hosting solutions. My proposal has no such limitations. If somebody manages to come up with a 500MB OXP, no problem! There are also no monthly bandwidth limitations. (even if said 500MB OXP gets downloaded 1000 times in one month, no charges are incurred)

The Oolite Wiki:
I have no idea if it is even possible, (at minimum it would need an upgrade of the Wiki software) but I'd like to see the Wiki and repository linked in such a way that updates at the repository are automatically reflected on the Wiki pages. Joomla calls such things 'bridges' and if it could be done, it would certainly fix one of the biggest gripes people have about the Wiki.

Update notification:
Many people have commented on how difficult it can be to stay on top of updates to OXPs, particularly when they have a large number installed. I propose to enable people to subscribe to receive automatic notification of updates to any or all OXPs. It would be easy for the repository to also track which OXPs subscribers have downloaded, if they wish, so that they would receive update notices only for the OXPs they have installed.

Death/boredom of Diziet Sma:
I do not intend to be the only one with full Admin capabilities. I want to have at least one (preferably two) senior community member(s) besides myself to have full Admin access. Once the system is up and running, update/addition changes take care of themselves. When an author makes a change, it will be reflected instantly, and notifications will be sent automatically. Most of the maintenance will be things such as occasional updates to the CMS software, OXP account creation for new authors (I don't want this automated, so as to prevent abuse by strangers) and adding new features.

Should it be deemed advisable, I would happily turn over ownership of the domain name to Ahruman (or whoever). The full offsite backups would enable one of the other Admins to move the repository to a different hosting company in the event of my death/disappearance. This is not an expensive thing to host, it's costing me just US$6 per month.

It is important to remember that there may be a "common" or "default" area for OXPs but it cannot, by definition, be exhaustive. There will always to works in progress (some never finished), private modifications, various version (with different textures perhaps) which may not all be public, some shared by email, and some licensed in a way that prevents centralising.

None of these really impact the planned scheme at all. WIPs, private mods and non-public versions will continue to be shared exactly the way they are now. In fact, for a (for example) WIP which is too large to host on box.net, such a central repository may be the easiest way to share the work with other team members/testers. Setting up a private area for projects like this (under the full control of the Author) would be easy to implement.

the theoretical oxp hosting site could be so very much more better than what we have now, no clutter, no download/upload size limitations, ability to make comments/screenshots, search functionality, top 100 oxp's, moderators, having an actual say in development of the site, not having to mooch off of alioth.net and box.net, + many more, are just some examples off the top of my head.

Yep, these are exactly the kind of things I'm trying to achieve.


Other business:
Drupal is back on my evaluation list, as it has undergone considerable development since the days of the Oosat2 crash.

[Diziet Sma]

re the larger file sized oxps.....

could some sort of peer to peer file sharing network be put in place?

i have a pc which is on on average maybe 6 hours a day which already has utorrent installed, i certainly would be happy to put aside a gb to share a torrent with a bunch of larger oxp's in.

so long as enough people were happy to do this, it would in theory keep the file alive.

i realise this has its own caveats with not up-to-date files but if a date was menioned in the torrent name then it should be simple to keep on top of it.

just a thought......

I like the idea, but in my experience, torrent sharing very easily uses a lot of bandwidth, and with the exception of the US with its' unlimited bandwidth packages, most people incur HUGE charges for exceeding their monthly 5/10/20GB allocation.

(I notice that ISPs in the US are currently reconsidering unlimited packages, largely due to torrent sharing negatively impacting profitability)

[Cmdr James]

I didnt mean it was a bad idea, actually I like it. I just meant that people seem to be jumping on this as a panacea when in fact it is a step on the road

[Diziet Sma]

I didnt mean it was a bad idea, actually I like it. I just meant that people seem to be jumping on this as a panacea when in fact it is a step on the road

Truly, I didn't take your comment that way. I was simply responding to points you raised. And yes, these are early steps, and it will take time to implement.

This is exactly why I posted this Request For Comment. I want to see what questions, concerns and ideas people have. I want to see what problems they anticipate. I want to see what features/capabilities people want. The clearer the picture we can build of what we want, the easier it will be to create it.

As ADCK has pointed out, we have the opportunity to make something that works the way the community wants it to work. The devs and OXP authors have created an amazingly impressive work in Oolite, and that work deserves to present an impressive face to the world.

My coding skills are too far out of date to easily contribute to Oolite itself, but this is something I can contribute to the community, to express my appreciation for both the wonder that is Oolite, and the community itself. This is for ALL OF YOU, and I want it to reflect your desires..

What features would you like to see in such a repository? How would you like it to work? As a player, and as an OXP author, what capabilities would make things easier for you? We are in the conceptual design phase at this stage, and I am trying to build a picture of what you want in your repository.

[Diziet Sma]

Thinking further on some of the above items, would OXP authors like to have private areas designed to make team collaboration on projects easier? It can be done.. think about what capabilities you'd like to have.

[Cmdr James]

One thing that would be nice would be a version control system (svn?). This would give visibility of changes, and give the potential to prevent loss of work if OXPs become abandoned (at least all source would be available).