Can I go to a project and immediately find a direct download link that I can paste into my terminal (for fetch or wget or whatever) ?
Or is it still user-hostile, requiring multiple clicks (pageviews) to get to a "download link" that gives me some 200 character long, ephemeral-link-monstrosity that assumes (wrongly) that I want to download their linux source tarball to my GOD DAMNED LOCAL MACINTOSH.
You know, I went to look at a new project on sourceforge, thinking I'd quickly prove you wrong. (I don't know why I still have any faith in sourceforge.)
Guess what happened? I clicked on TWO SEPARATE ADS with a download now image in them.
I eventually found the files link, and then clicked a download. And was taken to download page where I had to wait for my download to start, with some more ads.
Spoilers -- it is still plastered with ugly as sin ads, each one competing with the others in an arms race for your attention by plastering important looking 'buttons' (really just part of the ad's graphic) for you to click, making it difficult to figure out how to actually navigate the site.
Another thing that's missing in this overall thread is context....
.... context of companies like SF, Yahoo, etc. Basically, SF (Yahoo, etc.) has been in existence before the days of dropping server costs, dropping bandwidth costs, dropping storage space costs and before the term devops was even being used in conversations ;-)
Given this, and the fact that people were still figuring out how to scale and serve in the 90s, SF (along with Yahoo and some other web pioneers) was going ahead and practically doing it to scale, and doing it for "free" (if you train yourself to ignore the irrelevant ads, it is free!). To put this in context, Github/Bitbucket are all <5 years old by contrast. They've had all the time in the World to learn from the mistakes of these giants, and use all their learnings to get things right, and capitalise on EC2/AWS etc. (which was not possible during SF's setup period) :-)
So, given that some of these companies are the "giants" whose shoulders have enabled us to come this far, context is very inmportant in trying to understand these venerable beasts! They may appear to be long in the tooth by today's standards but they have lent us their shoulders, on whom today's jazz and pizzazz depend upon.
And for that I will always have a soft spot and gratitude for these internet pioneers like Yahoo, SF, etc.
EDIT: Also to remember is, the business model around file hosting/repository management was not the same in the 90s until mid-2000s as it is now, so that also needs to be kept in context.
I host my personal projects on Google Code. In contrast to Github it lets you host files (especially important if not all your users have compilers or similar tools), and lets you have multiple repositories per project. Pretty much every hoster does two repositories per project - one for the main source and a second for the wiki. Having more than two is great. For example for work stuff I often have Android client and server as different repositories. But there is no need for them to have separate bug trackers or wikis - for an end user it is often hard to tell where an issue actually lies and having to copy issues between client and server side trackers is useless work. Github's issue tracker lacks being able to prioritise which is highly annoying.
Having all our employees create new Github accounts and doing administration there is a pain when we already do everything else at Google. I suspect paid project hosting would also help improve Google Code since there would be revenue behind feature requests and bugs.
I don't understand what point you are trying to make. As I stated, pretty much every project hoster has two repositories per project for the code and wiki. Eg for github you have github.com/[user]/[project].git and github.com/[user]/[project].wiki.git
With Google Code you can have any number of additional repositories eg the equivalent of github.com/[user]/[project].documentation.git, github.com/[user]/[project].testsuites.git, github.com/[user]/[project].androidclient.git etc
With Github you would have to create new projects which means separate issue trackers, wikis etc.
Yes. I personally prefer hg over git and originally used bitbucket, including signing up a startup I worked at. The biggest problem was that bitbucket made it painful to have multiple accounts (work and personal) so I gave up using them for personal stuff. The startup is now defunct.
I did have some unpleasant experiences though, including a security hole (since fixed), which didn't leave me with much confidence.
Unbelievably sloppy, especially for something introducing a redesign. My browser is falling back to serif for most of the text... didn't _anyone_ at sourceforge test it on something not a mac?
SourceForge still has a niche that's significantly different from GitHub or Google Code: it's made to be somewhat friendly to end-users of large open source apps. Honestly, I'm surprised they haven't fully pivoted to being an open source "app store" (with more community features i.e. forums and support).
Yup. Especially after github abandoned the download feature which allowed developers to provide compiled binaries.
Github is great, but to end users which are just searching for good open source software to use its practically useless.
I hope sourceforge regains some of its old strength. I remember that when I was looking for open source software I googled something like "cd ripper sourceforge" and I would almost always find the current de facto standard open source solution to my problem. Without scanning through miles of useless entries filled with cdnet and the like.
Thus the "somewhat." Still, I've downloaded everything from torrent clients to open source video games from SourceForge, and many of those projects have (or, much more likely, had) active support forums on SF. It's a shame that they haven't pursued that route further and instead are just going after the Github/Bitbucket/etc market.
It's a lot better than github, IMO. Still not as good as a real project page with binaries, or ideally just having your package in macports, mac app store, ubuntu/debian repositories, etc.
I dunno. I would rather have a page with binaries plus macports, etc. There are a lot of cases where you may want to reach out to markets not covered by external repos or where you haven't gotten into every external repo you want to.
This app store idea also made me think that they need to show big screenshots of the apps immediately, and allow developers to put videos in there, too.
I think SourceForge was great but we've been spoiled by Github or Bitbucket. I never minded the ads, which are used to keep the site running, and were a minor inconvenient to provide access to a huge amount of open source projects, which was great.
Github and Bitbucket simply provided pages and sites which were easier to navigate. Specially Bitbucket, which has kept the Downloads section that Github removed. However, for me, the biggest "feature" that Github and Bitbucket provided was the ability to create projects easily because (a) the project name only had to be unique to your own projects instead of globally and (b) you could simply fill the project name and a brief project description to create it, instead of having to fill a form with lots of information about the project area and categories, licenses, development language or technologies, etc.
Note I haven't used SourceForge for new projects in a long time, so I don't know if my observations still stand true.
It's not quite the same as having native hosting of Subversion and Mercurial repositories. With a bridge and plug-in there might be some edge cases where things don't work as expected. I wouldn't risk my source code in this way.
By that logic github allows you to use CVS, bzr, darcs, monotone, TFS, whatever other VCS exists that has a git bridge of some sort. You miss out on any benefits of using that VCS over git and you introduce a huge performance sink.
Github's subversion bridge in particular is pretty second-rate - it's really only useful for IDE integrations which don't support git, and even then it's of questionable value. It falls over or provides incorrect data for almost any SVN request that falls outside the standard checkout-update-commit workflow.
There is also Gitorious for Git project hosting, and its open-source.
Unfortunately it shows the project's commit history when you visit a repository, and doesn't provide a way to show a description/README instead like all of the above hosting sites do.
I'm sorry guys at VA (Larry Augustine is an ass who sues people. Look up the OpenVista lawsuit for reference where he sued some of my friends) but if you want to be new and hip, buy github and dump the sourceforge brand. It was great once upon a time, but simply doesn't offer what developers are looking for. As a casual open source dev myself, if I saw a _new_ project using sourceforge, I'd hold it against them. Why? The sourceforge mailing list archive web ui is beyond terrible and dealing with the ads and terrible navigation make it waste my time. Sorry, but a pretty face won't save market share.
Dice Holdings (DHX) market cap is ~560m. GitHub is almost certainly worth more, growing like crazy, and generally almost finished with making SF irrelevant.
Man, when Sourceforge fixed the wget "misfeature" a few years ago, that was great. They should do more of that.
Every time I hit a Sourceforge page, entire seconds are spent navigating to the actual page that I want. That's not to say that it's impossible to figure out how to get to where I want (that status is reserved solely for Launchpad). Rather, it just takes a lot of time to use Sourceforge. When the seconds are ticking away, and another page has to load, it's just infuriating. I want at most two page loads: whatever awful incoming link I came in from, and then the thing I actually want on the project. I keep forgetting the url structure, so I can't just type to what I want to get. GitHub and BitBucket have both done urls particularly well.. you can easily predict the exact url to immediately see the contents of a file.
I know this is viewvc, but who cares. When I click a file, I wanted the file, not the version history. Why is version history still the default action? Okay, click on the version and you get the content:
At this point, I think a lot of Sourceforge's problem is viewvc. And the number of pages in between the project page and the things I need to do. I think getting rid of viewvc would probably be a huge step forward. Why should I even have to type viewvc in the first place?
When you don't include viewvc in the url, you get to a directory index (argh):
But why am I including phpmyadmin in the url again anyway? It's already in the subdomain. And it's not a folder in the repository... so what's the deal.
I agree completely ... but you also forgot the "sponsors". It's amazing that my download can't begin immediately, but that there's time to display an ad. I'm sure glad they didn't leave me there bored!
With LedgerSMB we have, in the past, provided direct links to downloads on Sourceforge. I do in fact use wget to download packages. It's quite possible to do that.
To be fair, it looks like those are all old-style projects. The new platform -- the one the headline links to -- sucks a lot less. e.g., http://sourceforge.net/projects/allura/. Source browsing works more like you would expect and less like ViewVC.
No, I don't know if they have a plan to forcibly upgrade old projects. I wish they would do so, because the new platform is far less terrible.
The project Allura page seems nice, altought I still don't understand why they have one sort of navigation in "code" view and in "non-code" view, that looks completely different. Also, the link "Pastebin" displays a 500 error in "non-code" view (http://sourceforge.net/p/allura/pastebin/?source=navbar) but works in "code" view ( http://sourceforge.net/p/allura/pastebin/ ) - you switch "views" by clicking at "Git", apparently. You switch them back by clicking at "Files" or "Summary".
Why does the Allura project have thousand of badly named branches? (It may be a goal but it looks confusing from outside.)
Now let's look at some other project that's linked at SF front page
What's the difference between "Wiki" and "MediaWiki"? Oh, once is inside SF, the other one directs me somewhere else and tells me it isn't working. Was that needed?
...why is the code part empty? Is that a bug?
The News are not very new but that's not an error on SF side, I guess.
Overall, I feel like SF jumped 5 years forward - from a site from 10 years ago to a site from 5 years ago. But I still wouldn't use it for any kind of new project.
It looks like there is some sort of plan in place as I received a "SourceForge Project Upgrade Notification" a few days ago letting me know that they've upgraded an old project of mine to the new platform. My project is small, really old, and unmaintained so maybe they're starting with low visibility/impact projects.
The worst thing is that most of these SourceForge URLs are loading horrible slow for me, sometimes producing 504 Gateway Timeout after 30 or more seconds. A few weeks ago i tried to read some code on the site but it was faster to download the source and browse it locally. GitHub's code browser works instantly like I would browse the code locally (except the annoying slide-in animation).
I am surprised that while so many of the comments are intensely hateful of the ads, but for a very tech-savvy site, most people actually seem to have no ad-blockers installed on their browser(s), which takes away virtually all of the ad problems? With ad-blockers, I find SF is just as smooth to navigate any any others out there.
I disliked and migrated away from SourceForge seven or eight years ago. Although it did let you provide compiled binaries / packages for end users, the process to do so was ludicrous, and the tools were paper-thin at best. rsync's comment about the download links are spot on; I can confirm that the Next SourceForge still gives you the ridiculous go-round when you want to grab the URL to an install package.
The catalog of software could be great - GitHub doesn't provide this, except in a developer-friendly format. And maybe this format still works (obviously I don't know their numbers). But it's slathered in ads and still reminds me of Tucows - I bet there's a better way to solve this problem.
Half joking, but for me, whenever I hear "sourceforge" - it's like a reminder from the 90s with heavy page loads and arcane user interface. It will take time to get over it.
I don't like to bash on other peoples' hard work, but it seems like their focus on features and not UX may be a mistake here. The reality is that most developers use other services now and they need to make a better case for why I should care about them. More features != more compelling.
Also I think the juxtaposition of the big green "Open Source" logo with "TM" is very ironic.
The redesign is an improvement, but the usability is still miserable. The main nav with "solution centers" is really just sponsors. The fake download banners and links are so misleading.
These attempts to get users to click on links makes my stomach turn. Compared to Github or Google Code, SourceForge still feels so dishonest. This is open-source software they are hosting here.
Nine years ago I was a subscriber of SourceForge, because I thought they were helpful to developers and the communities around projects. I'd never consider paying them anything these days.
I also found it surprising that Adium is the most popular Financial app.
I just visited the Allura project page to check out the new design, and my eyes had no idea where to look. There's just so much text with no clear layout. The job ads on the side have the same color and weight as the main project description, and there's so much extraneous information.
My head starts hurting every time I visit an SF page. Even after this redesign, I don't see why anybody would use SourceForge for a new project over GitHub + GitHub Pages for the project page, other than they really hate git.
Why is this site design giving me such intense deja vu? I swear I've seen a site that looked just like this before. Maybe it's just that all sites look kind of like this these days.
Eeh, yes and no. I think their ( = Open Source Initiative) main reason for doing so is to give them the right to pursue sites/companies that would use their logo while not actually being open source in some sort of attempt to gain goodwill with the people that would care about it.
It is not like your average open source advocates are standing on fences wanting the right to put an "endorsed by apple/google/amazon/..." logo on their sites.
How will that fix the problem at all? Will me changing the source code of my local copy of the source code somehow lead to the ads disappearing on SF, or me having to do less clicks when trying to download one of the six projects that still use SF for their hosting?
Can I go to a project and immediately find a direct download link that I can paste into my terminal (for fetch or wget or whatever) ?
Or is it still user-hostile, requiring multiple clicks (pageviews) to get to a "download link" that gives me some 200 character long, ephemeral-link-monstrosity that assumes (wrongly) that I want to download their linux source tarball to my GOD DAMNED LOCAL MACINTOSH.
I'm not even going to bother to look.