Live data from Hacker News

Open Source (Almost) Everything

tom.preston-werner.com

11–20 of 85 posts

Re: Open Source (Almost) Everything

#11
post #9

I have a question for the OP... Why the MIT licence, and not putting your code in the public domain? Is it the liability issues, or the attribution, or something else? I mainly pose this question because I am choosing a licence for my projects, either the MIT/Apache licence, or public domain.

Just a simple difference I know: Do you want to force users of your code to attribute the code to yourself? Then pick MIT/Apache

Re: Open Source (Almost) Everything

#12
post #9

I have a question for the OP... Why the MIT licence, and not putting your code in the public domain? Is it the liability issues, or the attribution, or something else? I mainly pose this question because I am choosing a licence for my projects, either the MIT/Apache licence, or public domain.

There is forced attribution and the fact that in some countries putting work under the public domain doesn't mean you give up a lot of rights to it. You need to have a specifically liberal licence in order to do that.

Re: Open Source (Almost) Everything

#13
post #8

Sometime I wonder if it is possible to open source everything without becoming some sort of service company, like Redhat. The only example I can think of is makerbot industries in which their core technologies are open source.

This is addressed in the article -

"Ok, then what shouldn't I open source? That's easy. Don't open source anything that represents core business value.

Here are some examples of what we don't open source and why:

Core GitHub Rails app (easier to sell when closed)

The Jobs Sinatra app (specially crafted integration with github.com)"

Re: Open Source (Almost) Everything

#14
post #9

I have a question for the OP... Why the MIT licence, and not putting your code in the public domain? Is it the liability issues, or the attribution, or something else? I mainly pose this question because I am choosing a licence for my projects, either the MIT/Apache licence, or public domain.

My understanding is that there are more legal gray areas with public domain than there are with even very permissive licenses (MIT, new-BSD, etc). I vaguely recall something about potential issues with public domain and other license incompatibilities, but I don't recall the specifics (and am certainly not a lawyer), so that probably isn't a useful datapoint.

Apache license has patent clauses that make it quite different than the new-BSD/MIT-style licenses.

In general, since I don't know a huge amount of the legal issues involved in public domain and national jurisdictions in regard to such, I simply err on the side of caution and choose MIT license for software I want to release as freely as possible. Others use it so it likely has more legal precedent and courtroom experience around it (should it ever come to that), and I like the notice of warranty disclaim (a bit of cya).

Re: Open Source (Almost) Everything

#15
post #9

I have a question for the OP... Why the MIT licence, and not putting your code in the public domain? Is it the liability issues, or the attribution, or something else? I mainly pose this question because I am choosing a licence for my projects, either the MIT/Apache licence, or public domain.

Just a simple difference I know: Do you want to force users of your code to attribute the code to yourself? Then pick MIT/Apache

A much better reason is that you can't easily put things in the public domain and many countries don't recognize the public domain.

Re: Open Source (Almost) Everything

#16
I've started open sourcing more things lately, mostly in the realm of things that solve a problem I have that might solve someone else's problem. Sometimes I build tech for tech's sake that improves my software or apps, but it isn't a product in itself. Frameworks, libraries, templates, and tools all tend to fall into this category I think and github is a great example is open sourcing these things.

You shouldn't open source your product itself unless you want to be a service company. I personally don't care to become a service company, so I keep the "product" and the data behind it closed, but the frameworks, tools, etc. can and should when possible be open source.

Re: Open Source (Almost) Everything

#17
post #8

Sometime I wonder if it is possible to open source everything without becoming some sort of service company, like Redhat. The only example I can think of is makerbot industries in which their core technologies are open source.

GitHub _is_ service company.

The primary advantage they have over competitors is a trusted brand name. This name has been built, in part, by engaging and contributing to the open source community.

If GitHub were to open source _all_ of their software, some copycat sites may appear. These sites would not have a trusted name, thousands of well known open source projects backing them up and the expertise to properly run an instance of GitHub. As long as GitHub doesn't try charging users too much to use their service, there is no reason for people to use competitors to save a few dollars a month (reliability is far more important than a few dollars here and there).

Some open source projects may decide to operate their own infrastructure. However, this is highly unlikely because GitHub _as a service_ is much easier and more reliable than an independently managed instance. If GitHub can retain the open source community, they become "the hub" that everyone wants to be part of. As long as GitHub is known to support open source in a meaningful way, given the resources available, I'm sure they'd benefit from hundreds of open source projects improving upon and patching GitHub code.

I think they could quite readily open source _all_ of their software.

Re: Open Source (Almost) Everything

#18
post #9

I have a question for the OP... Why the MIT licence, and not putting your code in the public domain? Is it the liability issues, or the attribution, or something else? I mainly pose this question because I am choosing a licence for my projects, either the MIT/Apache licence, or public domain.

Just a simple difference I know: Do you want to force users of your code to attribute the code to yourself? Then pick MIT/Apache

I'm no software license lawyer (and admittedly I haven't read Apache in depth), but I understand that the Apache license has at least one requirement I'm personally not too keen on.

4.2 in the license: "You must cause any modified files to carry prominent notices stating that You changed the files", I'm not sure how "legal" it is to follow the convention of sticking a LICENSE file in the project root and have that assume to cover all contents therein, but that seems reasonable enough, versus 4.2.

I'd love to know further opinions on MIT vs. Apache.

[edit: engaged brain.]

Post reply on HN