Google Code as a science repository

May 22, 2008

iPhylo is now available on Google Code. As Rod Page explains on his blog, inspired by Pedro, the goal is to use Google Code as a project management system, i.e. a place to store more than just code but docs and data as well (why not, you have good version control).

In essence this expands on the issue that I have been raising lately; that academics should use code repositories like Google Code, Sourceforge or Github. That not only moves some of the issues with code maintenance infrastructure and utilities out onto the cloud, it also brings in the ability of a bigger user base, ability to access mode more easily, etc.

Will this be a broader trend? There are already a number of packages available on Sourceforge, but still too few and many are just thrown out there.

Don’t know about all of you, but I find moves like these significant, especially if this becomes a trend

Update: As pointed out below in the comments, this discussion has taken on a life of it’s own on FriendFeed
. Disqus and/or FF really need to figure out a way to integrate those conversations

Technorati Tags: iPhylo, Roderic Page, Google Code

Posted by Deepak Singh | Filed Under Software & Internet

Amazon Recommends

On FriendFeed, this post was liked by 15 people and commented on 31 times show

View this post on FriendFeed

Liked by

May 22, 2008 at 11:03 am Pawel Szczesny

Is there any space limit for a project at Google Code? I've only read about limit in a total number of projects (ten if I remember correctly).
May 22, 2008 at 11:07 am PauloNuin

As my Firefox still does not like your site, I will comment here. I really don't understand all this buzz now, Sourceforge has been available for years and different scientific projects are available there, some for years now. Google Code is no different from Sourceforge, but I think Github is easier to use, faster and not difficult to learn and ultimately can be more widespread than the other two.
May 22, 2008 at 11:17 am Pierre

@nuin: I had an experience with both sites: IMHO Google is far more easier.
May 22, 2008 at 11:23 am Pierre

@pawel you may ask for more space: http://code.google.com/p/support/wiki/FAQ
May 22, 2008 at 11:27 am PauloNuin

Pierre, I host a couple of projects on Sourceforge and one on Google Code, and I don't fell one being more difficult or easier. Maybe GC's UI might be easier and at some extension default all over the board but Sourceforge allows more customization. IMHO, Github should be the ultimate choice.
May 22, 2008 at 11:28 am Michael Barton

I found Google Code's interface a bit unfriendly. There's nothing I can put my finger on specifically though. I thought one day that I might try and create my own version aimed at research projects rather than code projects. Just need to find the time.
May 22, 2008 at 11:33 am Pawel Szczesny

Thank you Pierre. I don't think I will hit the limits anytime soon, but it's good to know that overall they support extension of the quotas.
May 22, 2008 at 11:38 am Matt Wood

One idea that I think would add value: a central point of contact for scientific software. Scientific tools are usually housed on academic home pages, which makes tracking their updates, releases (and abandoment!) difficult. With a central repo, each project could be registered, and tools could be added to allow easy tracking, commenting, recommendations, paper citations etc.
May 22, 2008 at 11:40 am Morton Fox

I'm still very far from using up my 10 allotted projects at Google Code. Need to get more of my code up online.
May 22, 2008 at 12:15 pm Deepak

It's more about code repositories (pick your favorite one) that Google Code specifically. I definitely like the Github interface. Matt actually nails the issue on the head as far as my thoughts are concerned.
May 22, 2008 at 12:27 pm PauloNuin

So, who wants to work on a scientific software repository?
May 22, 2008 at 1:31 pm Pedro Beltrao

That could be included in the move to the cloud on the part of the big international databases (EBI/NCBI). Host projects and certify the stats in the third party independent manner. This could then be used when applying for further funding.
May 22, 2008 at 1:36 pm Matt Wood

It's certainly an idea that's been floating around in my mind for a bit: http://www.sciforge.net.
May 22, 2008 at 1:41 pm Matt Wood

@Pedro: Seconded. There are opportunities there for some kind of hosted application environment for science: a SciApp Engine, if you will.
May 22, 2008 at 1:41 pm PauloNuin

That would be nice, I would be willing to help ...
May 22, 2008 at 1:53 pm Michael Barton

Each paper would be required to have 100% test coverage to validate the code has been tested properly. I agree with the SciApp engine idea.
May 22, 2008 at 1:58 pm Pawel Szczesny

I'm not a good python coder (if at all), but count me in, if you get around to work on it.
May 22, 2008 at 2:12 pm Amund Tveit

Deepak: agree with you, using version control was number 1 (of 11) in my earlier posting about how to complete a PhD - http://amundblog.blogspot.com/2008/01/how-to-complete-your-phd.html
May 22, 2008 at 2:19 pm PauloNuin

http://blindscientist.genedrift.org/2008/04/23/how-to-improve-scientific-software/
May 22, 2008 at 2:26 pm Sally Church

By the way, as an aside if any of you science geeks would like an invite to Toluu for sharing science blogrolls let me know your email and I'll send an invite :)
May 22, 2008 at 11:51 pm Attila Csordas

A functional Google Palimpsest Beta would rearrange this conversation a bit, I assume.
May 23, 2008 at 12:37 am Neil Saunders

This talk of centralisation is rather "anti-cloud". I don't care where stuff is hosted, so long as I can find it. That said, Google Code makes most sense to me, simply because I view them as the future of large-scale data hosting. Maybe what we need is a portal/wiki page pointing to scientific code repos.
May 23, 2008 at 12:44 am Deepak

I am with Neil on the de-centralization part. I do like the idea of guidelines from publications on code repositories, much the way you do for structures etc
May 23, 2008 at 2:47 am Matt Wood

Neil: agreed. Somewhere to collate and store meta information about projects, rather than host them.
May 23, 2008 at 4:25 am Cameron Neylon

SciAppEngine - I see a scifoo session proposal coming on! But yes to efforts to aggregate the location of scientific software tools. Not to aggregate the applications, although hosting may be valuable in some cases, but to collect and index them.
May 23, 2008 at 4:39 am Pierre

Cameron: Pasteur's Biotnet Book started to create a list of bioinfo-softwares http://www1.pasteur.fr/recherche/BNB/bnb-en.html (there is also an XML version). I don't know if it's up-to-date.
May 23, 2008 at 5:30 am Michael Barton

I really like how big this discussion has become. I think a centralised repository for code would be great. I agree with the anti-cloud points though. Perhaps a better analogy would be a place where you can hook your code base into various feeds that broadcast interesting information when things happen. Similar to Github which allows you to hook your code into twitter and lighthouse.
May 23, 2008 at 5:39 am Matt Wood

Michael: that's exactly what I had in mind. A collection of tools to help aggregate code bases, services and applications around the web (other forges, personal web pages), rather than provide a centralised platform per se. Very meta.
May 23, 2008 at 9:02 am Cameron Neylon

Pierre: What I had in mind is more I think what Matt and Michael are talking about (they just put it better). But BioNetBook looks pretty cool as well. Presumably it is hand curated though? Or is there a wiki or crawler back end?
May 23, 2008 at 9:07 am PauloNuin

A meta-repository can be started by canvassing journals and current meta-repositories out there.
May 23, 2008 at 1:50 pm Pedro Beltrao

I can see the points about decentralization but at the very least an image of the code used for submission should be mirrored in a repository. The same standard for disclosure and long-term protection should be set to code as it is for scientific data.

Comments

Viewing 4 Comments

pansapiens 4 months ago 2 points
Please login to rate.

Do you already have an account? Log in and claim this comment.

It's amazing how the little differences in the interfaces between say, Google Code and Sourceforge seem to make participation easier. They essentially have similar features, and yet one always feels a whole lot more usable than the other. There are a lot of dead (or still-born) projects on Sourceforge, and I'm sure part of the reason is the messy setup for comments/bug tracking/documentation etc. It just doesn't seem as slick and easy to get an overview of what is going on in a Sourceforge project compared with one in Google Code (or Trac). I think we will see more tech-savvy (and hopefully the not-so-savvy) scientists realizing how nicely some projects can be managed this way and jumping on board in the future.

Ever since Neil wrote about setting up Trac + SVN to manage bioinformatics projects, I've been managing a few internal projects this way and it works great. Some of these will probably move to Google Code when I've cleaned them up for pubic consumption, but I really should get into the habit of just starting these out 'in the cloud' to begin with.

reply edit reblog flag record video comment

3 /people/pansapiens/ /people/pansapiens/following/ http://blog.pansapiens.com/

mza 4 months ago 2 points
Please login to rate.

Do you already have an account? Log in and claim this comment.

Some good comments over on FF: http://friendfeed.com/e/c610ba12-801e-0082-165c...

reply edit reblog flag record video comment

2 /people/mza/ /people/mza/following/

mndoci 4 months ago 1 point
Please login to rate.

Do you already have an account? Log in and claim this comment.

Matt

Thanks for posting the link. Can't wait for Disqus/FF to integrate those back into the blog. It's inevitable and Danial Ha mentioned they were working on that

reply edit reblog flag record video comment

1 /people/mndoci/ /people/mndoci/following/ http://mndoci.com

Pedro Beltrao 4 months ago 1 point
Please login to rate.

Do you already have an account? Log in and claim this comment.

That reminds me that I still did not make time to push on that open project ... hopefully when I get one project done in the new lab I will have an easier time convincing the boss that this is OK.

reply edit reblog flag record video comment

http://pbeltrao.blogspot.com

Google Code as a science repository

Comments

Add New Comment

Viewing 4 Comments

Add New Comment

Trackbacks

Lijit Search

Recent Posts

Comments

bbgm Recommends

Reader Map