Front page | perl.perl5.porters |
Postings from July 2010
On Sat, Jul 03, 2010 at 10:43:57PM +0200, Steffen Schwigon wrote: > There are big hypes that are taken over by companies. And they are > better done by companies because they are relatively boring in the > eyes of ???free??? people. Python, for instance, gets pushed via Google's > AppEngine. Its underlying topic is cloud computing, maybe irrelevant > in the eyes of non-corporate developers. In the end Google's > attendance turned into projects to improve Python's performance > (Unladen Swallow (Python on LLVM)). That's a valuable investment that > they receive. Which helps him independently from cloud computing. > Facebook provides a C++ backend to PHP. I don't know the details on Intel's Haskel contribution. I know that various firms pay people to write and maintain Java VMs, toolchains and libraries. Observe that a: Unladen Swallow and HipHop are not general maintenance work b: Both are contributions in the form of existing development teams working on code. Not monetary donations c: Both actually have significant trade offs, that mean that they are not general purpose drop-in replacements. Unladen Swallow uses a lot more memory than Python 2.6, and has a a start up penalty. HipHop only implements a subset of PHP 5.2 (not 5.3), and does not allow any dynamic code generation, such as eval > You will find such examples for lots of relevant language competitors > if you look for it. I did not find one for Perl. However, I find *no* relevant examples from anyone of how language competitors (Lua, PHP, Python, Ruby, TCL) have end user corporations sponsoring general maintenance of their original open source C implementations. Historically, of the many large commercial organisations using or still using large amounts of Perl, such as Shopzilla, Ticketmaster, booking.com, the BBC, none (to my knowledge) have donated developer time to the perl core. (Although I believe that Shopzilla, along obviously with New York Times, have contributed to the development of Devel::NYTProf) The only company I am aware of that ran a core infrastructure level project was Fotango, 5 years ago. Fotango was about 30 people in total. > As I said in another post: We should find a free language or compiler > that suffered from corporate support, to substantiate the risk. We have historically had code bombs from Novell (a poor quality monolithic patch for updated Netware support) and IBM (z/OS support, written by people who demonstrated a clear lack of understanding of EBCDIC vs ASCII). Both went nowhere. Corporate support has to be done in the right way. I read enough stories of Linux having commercial drivers dumped on it, that I believe that it's clear that we can't assume that all corporations know (or want) to do things right. That doesn't mean that none will do it right. But it's not a given. Nicholas ClarkThread Previous | Thread Next