Tech book publishers know that what they provide of value is access to a large reservoir of knowledge.
That doesn't mean they treat authors any better than other types of publishers. Most publishers severly undervalue their authors -- there's no way that the publisher provided seven times the value to my most recent few books than I did. (If they took on seven times more risk than I did, that's not my problem. That's their broken business model.)
Certainly he couldn't be prepared for every pitch, so how many pitches won on the basis of a strong (headed) arguing techniques instead of technical merit?
Yet people still think the Internet caught him by surprise.
Try having all of your accounts disabled in the morning and your boss calling you during breakfast to lay you off, then telling you to work remotely for the rest of the day.
You mean that Perl doesn't maintain a stable ABI from one version to another, at least for the parts that existed in the old version? What a bunch of amateurs!
That's not the problem in this case. By default, the shared library loader for XS components performs an exact version check against the Perl portion and the compiled portion, and refuses to load the latter unless the two match. Some combination of Apple's update plus how Apple ordered Perl's library include directory search order caused this problem. It has nothing to do with binary compatibility in the Perl core.
Perl 5 binary compatibility can change only between major version numbers -- that is, the 5.8.x releases are all binary compatible, but not necessarily binary compatible with the 5.10.x releases. (Breaking binary compatibility helped 5.10 have smaller and faster internal data structures.)
You could say the same thing about a D. Litt, at least in most places in the U.S., but it's still a doctoral degree. (No true Scotsman has one, though.)
One of the major points of the study is only the stopwatch knows for sure;/
I was in an airport several weeks ago, editing an interview that I had to submit in some sort of word processor format. I didn't have room to use an external mouse; I only had the trackpad. I spent ten minutes editing a few pages of the interview, increasing my frustration every time I had to move the cursor to a new place. I eventually exported the document to HTML, ran it through a beautifier script, and finished editing the remaining eighteen pages there.
It was faster for me to convert between formats and write some little Vim macros to perform global conversions and changes than to use an interface tied heavily to the mouse. Perhaps if I were better at scripting a word processor I could have made those changes there. I don't know.
I welcome the researcher with the stopwatch, however. Any UI which makes me feel like I'm wearing oven mitts with my hands on a keyboard behind my back is likely not faster. I'd like the chance either to discover that I'm wrong or to disprove this ridiculous claim.
I don't know that I can give specific details, but at 10% of wholesale, I don't make enough money in royalties to work as a full-time author of technical books. I believe there's still a market for the printed word, but I have my doubts that a big publisher creates as much value as it captures, at least from the author's point of view.
[We] can make money from advertising coupled with online references only if ours are consistently better than everyone else's, because there is no shortage of online references for computer software and languages.
To my knowledge, O'Reilly doesn't have advertising on its online references besides in-house ads. It's exceedingly difficult to make money from advertising unless you have hundreds of millions of page views a month. I doubt any technical online reference has that level of traffic.
That makes sense. While I oppose binary blobs philosophically and pragmatically, there's no way of determining at module load time whether they violated copyright. There's almost no way the kernel could prevent distasteful uses of binary modules while allowing less odious uses -- even if there were clear agreement about what's allowable and what's not. Preventing their use altogether seems incompatible with the ideals of the GPL and free software to me, let alone practically impossible -- hence my confusion!
They have absolutely no copyright over any of the code that nVidia provides....
No one's claiming that nVidia's source code is a derivative work of copyrighted Linux kernel code. Binary kernel modules are, however. (If you disagree, try compiling the module without the kernel headers. The resulting binary blob will be very different than one compiled against the kernel headers. Ergo the version compiled against the kernel headers contains information derived from the copyrighted headers -- unless you want to argue that a compiled version of copyrighted source code is not a derivative work of the source files, in which case I'm sure nVidia's legal department will quickly disagree.)
The kernel developers who signed that letter have copyright over their own code in the Linux kernel, which the binary blob includes. That's why nVidia distributes shim source code which, when compiled, links against both their binary blob and the kernel.
Nobody is forcing you to, so why do you want to force people to not use a binary blob?
I don't have the power to force anyone to do anything, and I don't recall ever attempting to do so. Are you confusing me with someone else?
Not only that, would you write a bunch of code in a language knowing that at any moment it could be invalidated by a few small tweaks? I wouldn't, not production code at least.
No one's suggesting you do so for Perl 6 right now. Ask again later this year.
But still until the language design is frozen, you will never be done.
Perl 5 is 14 years old, and its language design still isn't frozen. Almost every question of language design in the past year regarding Perl 6 has come from the implementors, whether people writing the language, people writing specification tests for the language, or writing applications in the language.
Really what is needed is someone like Larry Wall to finish his documentation, then someone to write tests based on the Perl 6 language design (In Perl 6) and then passing those tests can become a chart to Perl 6.
That's exactly what we're doing, just simultaneously.
If you're not advocating a waterfall-style approach I apologize -- but I get that impression. I've never seen that process work for any software, especially a programming language. I'm sure you can ask just about any other implementor of Perl 6 and receive the same answer.
Slashdot Mirror
That doesn't mean they treat authors any better than other types of publishers. Most publishers severly undervalue their authors -- there's no way that the publisher provided seven times the value to my most recent few books than I did. (If they took on seven times more risk than I did, that's not my problem. That's their broken business model.)
Good. Publishers have been cheating authors for years.
Yet people still think the Internet caught him by surprise.
Try having all of your accounts disabled in the morning and your boss calling you during breakfast to lay you off, then telling you to work remotely for the rest of the day.
That's not the problem in this case. By default, the shared library loader for XS components performs an exact version check against the Perl portion and the compiled portion, and refuses to load the latter unless the two match. Some combination of Apple's update plus how Apple ordered Perl's library include directory search order caused this problem. It has nothing to do with binary compatibility in the Perl core.
Perl 5 binary compatibility can change only between major version numbers -- that is, the 5.8.x releases are all binary compatible, but not necessarily binary compatible with the 5.10.x releases. (Breaking binary compatibility helped 5.10 have smaller and faster internal data structures.)
The cpan command is a thin wrapper around the CPAN module.
Indeed; can you provide citations that Wikipedia's aggregate quality has improved?
You could say the same thing about a D. Litt, at least in most places in the U.S., but it's still a doctoral degree. (No true Scotsman has one, though.)
The JD is a doctoral degree.
I'm eager to see Gallium too, but that's not the same as Apple dropping GCC for LLVM/Clang entirely for Snow Leopard.
Apple is, anyway.
What if I learned Scheme and Haskell?
Me too, but I'm not very good at C++.
I was in an airport several weeks ago, editing an interview that I had to submit in some sort of word processor format. I didn't have room to use an external mouse; I only had the trackpad. I spent ten minutes editing a few pages of the interview, increasing my frustration every time I had to move the cursor to a new place. I eventually exported the document to HTML, ran it through a beautifier script, and finished editing the remaining eighteen pages there.
It was faster for me to convert between formats and write some little Vim macros to perform global conversions and changes than to use an interface tied heavily to the mouse. Perhaps if I were better at scripting a word processor I could have made those changes there. I don't know.
I welcome the researcher with the stopwatch, however. Any UI which makes me feel like I'm wearing oven mitts with my hands on a keyboard behind my back is likely not faster. I'd like the chance either to discover that I'm wrong or to disprove this ridiculous claim.
I use Vim all day, almost every day. Using a mouse and a word processor is very much not faster for me.
I helped set up Onyx Neon Press for that purpose; it seems like a sustainable small business model.
I don't know that I can give specific details, but at 10% of wholesale, I don't make enough money in royalties to work as a full-time author of technical books. I believe there's still a market for the printed word, but I have my doubts that a big publisher creates as much value as it captures, at least from the author's point of view.
To my knowledge, O'Reilly doesn't have advertising on its online references besides in-house ads. It's exceedingly difficult to make money from advertising unless you have hundreds of millions of page views a month. I doubt any technical online reference has that level of traffic.
That's irrelevant in this case; O'Reilly is a privately held company.
That makes sense. While I oppose binary blobs philosophically and pragmatically, there's no way of determining at module load time whether they violated copyright. There's almost no way the kernel could prevent distasteful uses of binary modules while allowing less odious uses -- even if there were clear agreement about what's allowable and what's not. Preventing their use altogether seems incompatible with the ideals of the GPL and free software to me, let alone practically impossible -- hence my confusion!
No one's claiming that nVidia's source code is a derivative work of copyrighted Linux kernel code. Binary kernel modules are, however. (If you disagree, try compiling the module without the kernel headers. The resulting binary blob will be very different than one compiled against the kernel headers. Ergo the version compiled against the kernel headers contains information derived from the copyrighted headers -- unless you want to argue that a compiled version of copyrighted source code is not a derivative work of the source files, in which case I'm sure nVidia's legal department will quickly disagree.)
The kernel developers who signed that letter have copyright over their own code in the Linux kernel, which the binary blob includes. That's why nVidia distributes shim source code which, when compiled, links against both their binary blob and the kernel.
I don't have the power to force anyone to do anything, and I don't recall ever attempting to do so. Are you confusing me with someone else?
Distributing proprietary kernel drivers in compiled form does, however, violate the copyrights of many of the people who complained.
Perl 5's object system came from Python. See Larry Wall's Programming is Hard, Let's Go Scripting:
No one's suggesting you do so for Perl 6 right now. Ask again later this year.
Perl 5 is 14 years old, and its language design still isn't frozen. Almost every question of language design in the past year regarding Perl 6 has come from the implementors, whether people writing the language, people writing specification tests for the language, or writing applications in the language.
That's exactly what we're doing, just simultaneously.
If you're not advocating a waterfall-style approach I apologize -- but I get that impression. I've never seen that process work for any software, especially a programming language. I'm sure you can ask just about any other implementor of Perl 6 and receive the same answer.
The Rakudo spectest chart has daily updates of exactly that.