67 karma · joined July 12, 2010
"git cherry-pick" acquired the "--allow-empty-message" option
or
"git cherry-pick" now has the "--allow-empty-message" option
myself. I think the subject is fine with the emphasis where it belongs. The verb is weird, though.
http://www.scipy.org/Cookbook/Indexing#head-86055279f6592d36...
OMG
I guess server people would not encounter this problem.
Message passing is effectively the assembly language of distributed computing: very low level. Mapreduce and the various NoSQL databases are effectively the next level up, but we still have some time to be able to have a proper theory for designing and implementing effective scalable distributed systems.
In short, we have many bits and pieces that can be brought together with a bit of luck to be able to attain the scale that we are currently seem to be happy with. So I can accept that the engineering definition of solved is there. But from a computer science point of view, we certainly are far from having a satisfactory theory. So we definitely do not have the answers.
And the answer is .. 42?
No, really. When did the distributed computing issues become "solved" in computer science?
This is relevant: http://stackoverflow.com/q/3637668/336455 In short, there are some objective reasons, not just popularity.
I have been successful at learning a variety of languages. Ruby was difficult for me because it is so similar to Python and yet so different. You have almost the same syntax getting very different results. So if you come from Python you get the exact opposite experience to the principle of least surprise. On the other hand, if you move from OCaml or Erlang there are no surprises because the syntax is different enough that there are no pre-expectations.
Also, metaprogramming in Python and Ruby are quite distinct. The way Python went about it using metaclasses and decorators is very different to the monkeypatching you get in Ruby, so I don't expect the AST results to be that similar for idiomatic code in both languages.
For big companies to survive long term, the main thing is to have the flexibility to do that. The most relevant example here is IBM, which almost didn't survive the 80s until they emphasized more system integration.
As humanity progresses, demands will change. Microsoft has enough power, both in terms of number of employees and in cash, to pursue more than one option. Hence, many of the innovation coming from Microsoft Research can be very useful in the future. Moreover, one of the main advantages for Microsoft in the 80s and 90s was that they are friendly to developers. Of course there will eventually be a tie-in. You start using IronRuby, then you find ready .Net solutions to your programming, then you commit to using their database, and maybe visualisation via Excel, and you find yourself and your customers tied to the Windows/Office platform. Just like many companies are tied to IE6 despite what Microsoft wants and despite common sense.
Give developers no incentives at all, and they wander around. At some point, the clear division in tools makes people either commit to Microsoft or to everything else, and when Windows come short (and it will, nothing is forever) people will move in droves. But give them a semblance of cross compatibility, and they will stick to Windows/Office while convincing themselves that there is always a way out. In reality, they will be locked by the economics of rewriting all the use of .Net technologies.
Bad for the community, because splitting the community into Microsoft and others camps will lessen competition in both sides. Cross pollination will C# and Ruby will not exist when there is no shared community. And no improvements will be needed in either camp because of new features in the other.