Software-as-service (SaaS) has started to permeate the field of software development tools, says Atlassian's Jeffrey Walker in this interview with Artima:

http://www.artima.com/lejava/articles/javaone_2008_jeffrey_walker.html

What do you think of hosted development tools?

> What do you think of hosted development tools?

The same thing I think of hosted Office software: junk. There are some niche markets for such things, but outside those narrow niches, they are junk.

Which niches do I see?

1. Creative games that allow you to script them live. (Something like Second Life) A player is a developer and a developer is a player paradigm on a multi-player game. Neverwinter Nights has already enabled a player is a developer paradigm without it being hosted. So it's a niche within a niche. Within a niche of "player is a developer" there is even a tinier niche of multi-player games where players may want to develop things.

2. Hosted development that's embedded in some highly creative collaborative environments. Think Wiki where you can edit not only text, but the code too. Again, this is a tiny niche. I can see such code-editable Wikis being popular with open-source developers. I don't see it getting traction beyond that.

That's about it.

The general trend is away from centralization, not toward it, especially in the development community. Think of version control software. SVN's popularity is stagnating and within the open-source crowd is on the way down. Interest in DVCS is going up and up. People want to engage in highly decentralized coding where everyone is a king of one's own little kingdom while at the same time enabling great inter-kingdom collaboration. DVCS precisely enables this. If people want to develop collaboratively, there are always great ways of doing it that hosted development will never match 1 to 1. For example, there is a collaborative coding plug-in in NetBeans. The trend I see there is peer-to-peer rather than hosted.

And that's another thing. Peer-to-peer protocols are getting more popular. Hosted protocols are declining. People generally don't like the idea of a single point of failure. And that's exactly what a hosted development environment is. How is hosted better than p2p? You get all the networking benefits from p2p without feeling like some corporation is in control of your data and like you depend on a single point of contact, etc.

I have 2 or 3 Google Office docs. They contain junk. They contain throw away data. Would I store anything truly worthwhile on Google? Would I put any kind of crown jewels on Google? Heck no. If you notice what it's used for, it's mostly to quickly share some data that's public anyway. No one uses Google Office seriously.

Hosted computing just isn't going to dominate. People enjoy peer-to-peer. People like the flexibility of working offline. People don't like single point of failure.

These "tiny" niches might have millions of people in them and might be quite profitable, but I just don't think any kind of hype is warranted there.

That's my perception of the situation. Just my 2c.

I see it differently. I use Google Apps for all my business documents, presentations, invoices. It's worth it to me for the consistent interface on all platforms and the default backup. Would I pay for it? Yes, probably up to $10/month.

As for hosted development environments, back in 2005 in "Practical Development Environments" (http://www.oreilly.com/catalog/practicalde) I used the term "preconstructed development environments" (PDE) for this, and wrote:

"Before you outsource your development environment, make sure that you have good answers about how much you can customize the environment, how secure it is, how backups are
performed and tested, and how you would extract your project from the environment to move it somewhere else."

and also:

"It's one thing to use a remote preconstructed environment such as SourceForge, where you're depending upon a remote group of toolsmiths to develop and maintain your development
environment for you. It's quite another to expect to be able to run a single installation command and have a working SCM tool, a bug tracking tool, a web server, and so on. No PDE
that I'm aware of is there yet, but I think that this is where PDEs should be heading."


Maybe virtual appliances will get us there.

Leo, it seems to me that your comments about Subversion are valid only for open source, distributed development. Far from stagnation, I'm seeing a large adoption rate within organizations that want to retain control over their source. The jury's not out on this one yet.

> Leo, it seems to me that your comments about Subversion
> are valid only for open source, distributed development.
> Far from stagnation, I'm seeing a large adoption rate
> within organizations that want to retain control over
> their source. The jury's not out on this one yet.

Sure, I agree. The jury can preside for as long as they like. But I'm out. :) I see a trend toward DVCS and toward decentralization. It's not something tiny. It is a huge tidal wave, effects of which are so staggering that ISPs are resorting to legal gimmicks, lobbying, cheating the user, and other dirty games, to push back against this irreversible tsunami.

When I see someone offering a hosted service of any kind, I see it as a step backward in time.

As for the Subversion, the only, and I mean the only reason why we use it at this large corp., is because it has a nice Windows GUI (TortoiseSVN) and because it can keep its repo on a shared drive. Period. How do I know this? Well, I was the decision maker on that one. :) As soon as a DVCS appears with a nice GUI, it's game over for Subversion. Shared drives won't be relevant anymore either. If everyone can keep a repo on their C:\ drive, it's all gravy. And you get a natural backup, because all your developers have a full copy of the repo with all the histories. This is the peer-to-peer benefit.

I think Subversion will remain in use for the same reason MySQL remains in use, even though it is drastically inferior to PostgreSQL -- inertia. Except I don't think inertia will be as kind to SVN as it has been to MySQL.

SVN was and is a great tool. SVN was necessary at the time. SVN is probably still necessary as of yet. But I do see the future going toward decentralization, even within the corporate environments. Corp. users want ease of use, but they do understand the benefits of decentralization. If you can give us the decentralized p2p approach that doesn't sacrifice ease-of-use, you bet we'll be using it.

Matt, one more thing. While Google has great up time it is not 100%. I've hit Google at the times when it was down. On the other hand, I have multiple computers at home and work and multiple backups/copies of the necessary files, and I've never encountered a situation where all my computers and storage went offline simultaneously.

I've also encountered moments of spotty network connectivity. While ISPs have majorly improved network stability since the 1990's, once in a while I still experience laggy or spotty connectivity somewhere closer to the backbone.

Finally, in my experience Google apps are noticeably sluggish and not exactly feature-rich. Sluggishness will improve as the new browsers come out (FF3 makes huge performance gains over FF2) and features can be added (potentially bringing the sluggishness back even on the new browsers), but then desktop apps don't stand still either.

Even if all this is overcome, the privacy issues will be a deal breaker. Does Google assume liability against data theft? If someone steals your data from Google, can you sue it? Can you stop Google from giving your data to the government agencies without a court order or from giving your data to your competitor? Does Google have underpaid and/or disenfranchised employees who could be bribed to give your data to your competitor easily?

Storing data on your own computer is no guarantee of security. But at least there you are in control. Being in control may be of no value to people who don't know what they are doing, but it is valuable to those who do know. Also, if everyone stores their data on personal computers, the attack surface area is huge. On the other hand, if many people store their data at once place -- Google, that narrows the attack surface area enough to make it more attractive to try to break into it and steal the data.

> Storing data on your own computer is no guarantee of
> security. But at least there you are in control. Being
> in control may be of no value to people who don't know
> what they are doing, but it is valuable to those who do
> know.

I don't want to think about how small of a percentage this is. As long as we're talking about trends, this percentage is heading steadily towards vanishingly small. Computer security is not getting any simpler. More importantly, people *don't want* to care about.

Speaking for myself, I don't want control of my security. I don't have the time and I don't have the energy. Besides, I'm atrociously bad at it.

Being in control is *anti*value to people who don't know what they're doing. If you cede control, at least you have someone to blame. (You can't sue yourself, either ;)

I provide consulting services for version control, build systems, bug trackers and development tools in general. You can replace "security" with any one of these tool areas in your comment and you've got 80% of my customers.

I always tell nervous customers to ask me to make the changes so that they can blame me later on. It's all part of the service provided by a good consultant :-)

~Matt

Can somebody be so kind to fork out some bucks and send Scott to a hairdresser. Everybody in IT complaints about not being taken serious enough by the business department's inhabitents ... I WONDER WHY THAT IS?!

Cheers,
Stan