Home
 
Previous | Memories | Tell a Friend | Next
 
kernelslacker
kernelslacker
08 June 2007 @ 05:45 pm
why sourceforge sucks. Episode 37.   
In accordance with our data retention policy, this project does not
qualify for removal.  I suggest that you update the public information for
the project to indicate that development has moved elsewhere and perhaps
update your project's webpage to the new (relatively) development site.

Based on a brief review of your project, it appears that your project has
produced a body of work which may be of use to other developers in the
future. SourceForge.net provides hosting to a variety of Open Source
software development projects, and projects producing documentation for the
Open Source community.  It is the policy of SourceForge.net that projects
which have produced material shall not be removed except under special
circumstances (such as some legally compelling reason, i.e. DMCA or
trademark issue).


Yet another reason why decentralised source control is a good thing.
 
16 comments | Leave a comment
 
( Post a new comment )
Hubert Figuiere[info]hub_ on June 8th, 2007 09:59 pm (UTC)
I often mistype sourceforget.net. I know why :-)
(Reply) (Thread) (Link)
[info]srlamb on June 8th, 2007 11:01 pm (UTC)
I don't get it
They don't want to remove anything potentially useful. Assuming there is a decent way to direct people to the newer location, why does that suck, and how would decentralized version control have solved it?
(Reply) (Thread) (Link)
kernelslacker[info]kernelslacker on June 8th, 2007 11:14 pm (UTC)
Re: I don't get it
google for x86info. Note which hit gets returned first.
Because theres no way to remove the project, this will probably stay there forever now due to the googlejuice that sourceforge has.

Yes, I can redirect people from that page, but the redirection text is one tiny line lost amongst the various advertisements for sourceforges 'services' and other crap.

decentralised version control solves the problem of 'we dont want the only copy of this repository to ever vanish' because everyone has a copy of the whole repository. The git tree for x86info has the entire CVS history of the sourceforge CVS repo, so there's no value at all in having the old branch around, especially when it's the first thing people find when they google for it.
(Reply) (Parent) (Thread) (Link)
[info]srlamb on June 8th, 2007 11:21 pm (UTC)
Re: I don't get it
Ahh. So there's not a decent way to redirect people to the new page. Yeah, that sucks.

But decentralized version control doesn't guarantee the project lives on forever. If people already have a copy of the repository, they still have it even if yours goes away, yes. But it doesn't guarantee that there will always be at least one publically-available repository, and that's what they claim to be concerned with.
(Reply) (Parent) (Thread) (Link)
smitty1e[info]smitty1e on June 9th, 2007 12:39 am (UTC)
Re: I don't get it
At some point, time will be wasted trying to figure out the authoritative copy.
Someone will also mistakenly overwrite good code with bad.
Price of freedom, I suppose.
(Reply) (Parent) (Thread) (Link)
Joe: fart[info]mr_z on June 9th, 2007 01:43 am (UTC)
Re: I don't get it
I guess you could be obnoxious and replicate your "Nothing to see here; come over here instead" text until it takes up way more screen real estate than the crap. Amplify the signal until it drowns out the noise.

But, then, that would be, by definition, obnoxious. :-P
(Reply) (Parent) (Thread) (Link)
Speaks With Fingers[info]technoshaman on June 10th, 2007 05:07 pm (UTC)
Re: I don't get it
One could further make things interesting by not only being very loud about "these are not the droids you're looking for" but saying "and by the way, Sourceforge sucks and here's why" on said summary page... and you could also generate a package update that, when you try to install the package, errors out with messages to the same effect.

I don't like them either; they bother me for much the same reasons. Such things should not be run for profit; the customers inevitably get screwed.
(Reply) (Parent) (Thread) (Link)
[info]jamezm on June 9th, 2007 06:21 am (UTC)
I wonder if copyright law can help you here.
(Reply) (Thread) (Link)
[info]jamezm on June 9th, 2007 06:32 am (UTC)
Drops of google juice
re. x86info,

http://namei.org/misc/x86info.html


(Reply) (Parent) (Thread) (Link)
Ian Monroe[info]eeanm on June 12th, 2007 12:21 am (UTC)
This isn't so much a statement of centralized vs. decentralized source control. Mostly its that Sourceforge (while it does provide or at least used to provide a valuable service) sucks.

I mean... you still need to stick your website somewhere.
(Reply) (Thread) (Link)
kernelslacker[info]kernelslacker on June 12th, 2007 12:47 am (UTC)
suppose I go crazy and decide to pull the repository off my web page.
Anyone who has an existing checkout has the complete history.
They have everything that I have.
Sure, I could also make the raw bits of a CVS repo available, but that would be a bit nuts. With a complete local repository, anyone could pick up where I left off.

But yes, sourceforge does suck, and not just because of its continued use of centralised VCS.

A better written post would have mentioned how I was more pissed at losing control over the repository. Once you've given sourceforge 'bits', they basically take ownership.
(Reply) (Parent) (Thread) (Link)
Ian Monroe[info]eeanm on June 12th, 2007 01:01 am (UTC)
I'm a developer for Amarok. It started out on Sourceforge and we gradually moved it off to KDE.org (first the VCS, then the website, then the mailing list and just last year the tar balls). Cause yea, their website really does suck. Its just filled with too much crap and is ugly. I remember their anonymous CVS server was down for a while this last year. I was really glad to see Google Code start, since there is a need for a free easy-to-start open source website/VCS/bug tracker solution.

The funny thing is that apparently the admin of the sourceforge group at least used to get periodic emails complaining about how all the tarballs overfilled the quota and that they would be deleting it. I just looked and they are still there.
(Reply) (Parent) (Thread) (Link)
kernelslacker[info]kernelslacker on June 12th, 2007 01:18 am (UTC)
Yep. The project page is now just an advertisement for sourceforge services, with a lost-in-the-noise "oh, and btw, what you were looking for moved".

It's the primary thing that should be visible on the page.
Hell, it shouldn't even be visible at all, what's wrong with it setting up a forwarder to the page I told it ?

I notice that amarok's old sourceforge page doesn't appear in the first page of google hits at all, so there's some hope yet.
(Reply) (Parent) (Thread) (Link)
Paul[info]arafel2 on June 15th, 2007 02:01 pm (UTC)
> Anyone who has an existing checkout has the complete history.

That's assuming anyone has a checkout. If they don't, then the project's lost, and that's what they're trying to avoid. Even if $random_person has a checkout sitting unused on their disk, the chances of $other_random_person getting hold of that if they want to continue develop is minimal. Basically, this is not one of the problems which distributed SCM solves.

To be honest, I think what they're trying to do is a good thing. Where I agree with you is that they should make it much, much more obvious on their pages if the active work on a project has migrated elsewhere.
(Reply) (Parent) (Thread) (Link)
Hawke[info]hawke666 on June 14th, 2007 03:41 pm (UTC)
Why not put something at http://x86info.sourceforge.net/ that simply redirects to the new page?

I would think that URL would be sufficiently high up in the search results, and would send people to the new site as necessary, without risk of being lost in the noise of sf.net ads & such.
(Reply) (Thread) (Link)
kernelslacker[info]kernelslacker on June 14th, 2007 03:53 pm (UTC)
hmm, probably a good idea. thanks.
(Reply) (Parent) (Thread) (Link)