Interview- Guido van Rossum - The Modern Era of Python

Python was created by Guido van Rossum in 1989 while working at CWI in the Netherlands. Python was released as Open Source in February. 1991. The open source community grew around the world. And the first Python workshop was hosted by NIST in 1994. From 95 to 2000, I worked in the U.S. in Northern Virginia at CNRI. And so, there we worked through a lot of growth of the community and the infrastructure. We created the Python website. I think when I started there, Python 1.3 was about to be released. And then while I was there, we released several subsequent versions leading up to 1.5.2. Which for some reason, 1.5.0 was nothing, but 1.5.2 remained the sort of, the standard, the gold standard of Python for a very long time afterwards. Disagreement was brewing between the people working on Python at CNRI and the CNRI leadership about open source. I was sort of courted by a small no name startup, and they said, oh, you gotta come work for us. We’re an open source startup. We’re going to build an open source portal. And find a number of other good Python engineers, and we’ll give you all the time you want to sort of work on Python full time. And I thought, wow, that sounds great! Especially, given that I already had this sort of discontent at the place where I had been for five years, and so, reluctantly, I agreed to join these guys. Well, by then, by the time we actually started, it was 2000. And with, our start date was May 15, the bubble had actually already burst. I wasn’t even aware that there was a bubble and that it had burst. I was in Northern Virginia doing open source software. I wasn’t, I didn’t really know what a startup was. I didn’t really know how to tell whether these people were trustworthy or not. Well, we spent the summer in blissful ignorance working full time on Python. We built and released Python 2, which was a big deal. It contained Unicode. We also, because we had to have some kind of graceful exit from my previous employer, who claimed a certain amount of ownership over the source code. They didn’t claim that it was all theirs, because it wasn’t theirs to begin with. It was open source before I joined, which was an incredibly lucky circumstance. But, nevertheless, they sort of claimed ownership over what had been added to it over the five years I worked there and other people who were also employed there, also worked there. And so, with negotiations facilitated by Eric Raymond, and I think also even more in the background Eben Moglen, the free software lawyer, we negotiated some way to make sure that Python remained open source, with a license that CNRI’s lawyers were agreeable to. And that was sort of, Python still has a very awkward license. And the history of that awkward license lies in the way I left CNRI. So, this little startup, we didn’t even move to the west coast. We just sort of, we worked from our homes. And I remember that once a week, we had a team meeting and a catch-up with the leadership in California in my living room. And within five months, it was over. They suddenly stopped paying us. And those five months, we had had the time of our lives working on Python, doing this Python 2 release. We also had to do a Python 1.6 release, that was actually nearly identical to Python 2.0. That was the first Python 1.0 release with CNRI’s license on it. And sort of over the summer and the fall of 2000, this startup company began to show more and more signs of dysfunction. Other teams suddenly disappeared. We later heard that they had been fired. They flew us out to the west coast to talk to random potential customers, to sort of show off our technical prowess and somehow, get some kind of deal where we would do work for a customer. None of those deals ever closed. So they were going deeper and deeper in the hole. And at the end of the year, the investors decided to stop funding it. And then suddenly, we were like, we were out in the streets. We were relatively well off, because we had been paid well until then. But, yeah, the company was gone. So, what, all we got out of it was, they had given each of us, they had bought each of us a good computer to work from home. So, that was our last payment. Cuz we talked to lawyers or somehow we got advice where someone said, yeah, sure. Just hold onto that equipment. Nobody’s going to have the power to take that away from you. They’re not going to go after you. And so, then there was actually, well I wouldn’t call it a bidding war. But we were negotiating with both Active State, which was a small software company in Vancouver, and Zope Corporation, which was a small Python-exclusive software company in Northern Virginia. And the whole Python labs team, five of us, ended up working for Zope. And that was an incredibly lucky rescue. With the terms and conditions that pretty much keep working on Python, right? I mean, that Zope didn’t tell you you had to go- No, Zope was very clear. They said, we have no design on the ownership of that software. We want you guys, you guys are really good programmers. You guys are the top of the Python community. You are going to get some time and it wasn’t stated very precisely how much time. And we were fine with that at the time. We realized that what we had done for this failing startup was not actually a realistic option. That they would just pay our salaries and we would only do open source development. Nobody could maintain that, so we had sort of grown up quickly. We did continue to do a lot of sort of fundamental Python work at the time at Zope. We also did a lot of fundamental Zope work. Over time, sort of, the team dispersed. I got a job offer to come work for a small startup again in California. This time I did my homework a little better. Python just kept, sort of, growing, and the community kept self-organizing. And so, what started out as a workshop we had with about 20, 25 people at most attending in ‘94. Within a few years, we had an annual, what we called the International Python Conference, with with three or four hundred people attending. I always encouraged the community to self-organize rather than sort of looking at me for every decision. And so, one of the things that gradually appeared was the notion of Python warts. I’m not sure, but I think that Andrew Kuchling was actually one of the first to talk about Python warts, and sort of blog about it or give a talk about it. Of course, different people have different ideas about what’s wrong with Python. But, sort of the idea of warts is things that keep tripping people up over and over. And so, there is at least some notion of, these are objectively problems. Every sort of feature release, 2.1, 2.2, 2.3. We tried to fix some of these things. But we also came up against a wall of things that we could never fix because the fix was sort of, required changing the language in an incompatible way. Because it’s always easy to add new syntax, that sort of doesn’t get in the way of existing syntax. But it’s very hard to take away a particular syntactic construct that has an unfortunate but well-defined meaning. And all these things built up and, sort of, in the middle of the first decade of the new century, this idea of Python 3000 was born. Like, okay, we have so many things that we would like to fix. We know roughly how to fix them, but the fix would be incompatible. Let’s create a new version of the language that fixes a whole bunch of those things together. While at the same time, not making the new version of the language so different that it alienates the users. Every year at PyCon, I sort of take the census of well, who is using Python 3 and how committed are they? And how happy are they? And sort of, are there users actually using Python 3? And, I see tremendous progress there every year. But, we’re still not, we’re not in, sort of, where we finally want to be. I mean, at the last PyCon conference, I had to, sort of, hold off a whole bunch of people with very good arguments of why we should do a Python 2.8 release. And so, Python 3 is winning the race. But it will, the race hasn’t been run yet. We’re still in the race. It’s clear that Python 3 is going to win it.