Embed Size (px)
Citation preview
c© 2013 Andrei Alexandrescu 1 / 46
Quo Vadis?
Andrei Alexandrescu
Research Scientist
aspirations
c© 2013 Andrei Alexandrescu 2 / 46
Principled & Practical
c© 2013 Andrei Alexandrescu 3 / 46
• Slices
• Approach to modularity
• Support for generic programming
• Error handling done right: scope
• Qualifiers
• Approach to @safety
“Meh” features
c© 2013 Andrei Alexandrescu 4 / 46
• @property
• synchronized’s interplay with shared
• delete
• foreach_reverse
• qualified postblit
Features that don’t exist (but work)
c© 2013 Andrei Alexandrescu 5 / 46
• Attribute inference
• Compile-Time Function Evaluation
• Scoped imports
• Value Range Propagation (no-pain conversions)
• Relaxed purity
Functional Factorial (yawn)
c© 2013 Andrei Alexandrescu 6 / 46
ulong factorial(uint n) {
return n <= 1 ? 1 : n * factorial(n - 1);
}
• It’s PSPACE!
• Somebody should do hard time for this
However, it’s pure
c© 2013 Andrei Alexandrescu 7 / 46
pure ulong factorial(uint n) {
return n <= 1 ? 1 : n * factorial(n - 1);
}
• Pure is good
Functional Factorial, Fixed
c© 2013 Andrei Alexandrescu 8 / 46
pure ulong factorial(uint n) {
ulong crutch(uint n, ulong result) {
return n <= 1
? result
: crutch(n - 1, n * result);
}
return crutch(n, 1);
}
• Threads state through as parameters
• You know what? I don’t care for it
Honest Factorial
c© 2013 Andrei Alexandrescu 9 / 46
ulong factorial(uint n) {
ulong result = 1;
for (uint i = 2; i <= n; ++i) {
result *= i;
}
return result;
}
• But no longer pure!
• Well allow me to retort
Pure is as pure does
c© 2013 Andrei Alexandrescu 11 / 46
• “Pure functions always return the same result for the
same arguments”
• No reading and writing of global variables
◦ (Global immutables okay)
• No calling of impure functions
• Who said anything about local, transient state inside the
function?
Transitive State
c© 2013 Andrei Alexandrescu 12 / 46
pure void reverse(T)(T[] a) {
foreach (i; 0 .. data.length / 2) {
swap(data[i], data[$ - i - 1]);
}
}
• Possibility: disallow
• More useful: relaxed rule
• Operate with transitive closure of state reachable
through parameter
• Not functional pure, but an interesting superset
• No need for another annotation, it’s all in the signature!
User-defined types
c© 2013 Andrei Alexandrescu 13 / 46
pure BigInt factorial(uint n) {
BigInt result = 1;
foreach (i; 1 .. n + 1) {
result *= i;
}
return result;
}
• Works, but not in released version
• Not all stdlib “purified” yet
Aftermath
c© 2013 Andrei Alexandrescu 14 / 46
• If parameters reach mutable state:
◦ Relaxed pure—no globals, no I/O, no impure calls
• If parameters can’t reach mutable state:
◦ “Haskell-grade” observed purity
◦ Yet imperative implementation possible
◦ As long as it’s local only
c© 2013 Andrei Alexandrescu 15 / 46
comprehensive != big
c© 2013 Andrei Alexandrescu 16 / 46
comprehensive != perfect
Allocators
c© 2013 Andrei Alexandrescu 17 / 46
• We want a comprehensive design
• (not a big/complex/perfect one)
• It should work
Allocation Archetypes
c© 2013 Andrei Alexandrescu 18 / 46
• Garbage collected
• Garbage collected + free
• malloc-based
• Region-based
Composability
c© 2013 Andrei Alexandrescu 19 / 46
• Allocators must stack efficiently
• Example: freelist over region
• (Related work: HeapLayers)
Safety
c© 2013 Andrei Alexandrescu 20 / 46
• Some allocators safe (GC)
• Some unsafe (malloc)
• Some can be made safe (regions)
vision
c© 2013 Andrei Alexandrescu 22 / 46
Fable
c© 2013 Andrei Alexandrescu 23 / 46
Neophyte: “What should I do to be a
better person?”
Fable
c© 2013 Andrei Alexandrescu 24 / 46
Guru: “Do what a good person does.”
Applied
c© 2013 Andrei Alexandrescu 25 / 46
We: “What should we do to scale to 1M
users?”
That means
c© 2013 Andrei Alexandrescu 26 / 46
• Stability
• Quality, quality, quality
• Expanding platform base
• Operational professionalism
c© 2013 Andrei Alexandrescu 27 / 46
Stability
c© 2013 Andrei Alexandrescu 28 / 46
Quality
c© 2013 Andrei Alexandrescu 29 / 46
Expanding platform base
c© 2013 Andrei Alexandrescu 30 / 46
Operational professionalism
Two-pronged Approach
c© 2013 Andrei Alexandrescu 31 / 46
• Play into strengths
• Improve on weaknesses
Strategic strengths
c© 2013 Andrei Alexandrescu 32 / 46
• Active community
• Incredibly fast turnaround
• Compile-Time Function Execution
• Domain-Specific Embedded Languages
• Libraries
• Ranges and algorithms; bulk processing
• Concurrency and parallelism
• One-stop shop for getting work done
Weaknesses
c© 2013 Andrei Alexandrescu 34 / 46
• Quality of implementation
• Formal definition
• Available libraries; package management
• Ecosystem tools
• Documentation and tutorials
• Process and roadmap
people
c© 2013 Andrei Alexandrescu 35 / 46
c© 2013 Andrei Alexandrescu 38 / 46
Make DConf an annual
rallying point
The People Connection
c© 2013 Andrei Alexandrescu 39 / 46
• Any community needs nurturing
• Focus on increasing participation
• Welcoming new community members
• D Summer of Code (DSOC)?
• forum.dlang.org dedicated discussions place
“D is totally useless”
c© 2013 Andrei Alexandrescu 40 / 46
“When i tried to rewrite example to D, i was
shocked. [...] Dlang is a toy in outer space. [...]
One can only to write a+b program in schools in it.
Now I understand, that’s why D doesn’t have
popularity after 10+ years of existence.”
– Temtaime (tinyurl.com/d-useless)
On second thought. . .
c© 2013 Andrei Alexandrescu 41 / 46
“I investigated a little more in it. Thanks to Jack
Applegame, we made a copy of gl/gl.h and
opengl32.lib for DMD.
(http://acomirei.ru/u/gl.d,
http://acomirei.ru/u/opengl32.lib). I
hope it will be included in DMD, now it’s first draft
of our work.”
– Temtaime
c© 2013 Andrei Alexandrescu 42 / 46
Community
c© 2013 Andrei Alexandrescu 43 / 46
Academics
c© 2013 Andrei Alexandrescu 44 / 46
Corporate
Summary
c© 2013 Andrei Alexandrescu 45 / 46
Summary
c© 2013 Andrei Alexandrescu 46 / 46
aspirations
Summary
c© 2013 Andrei Alexandrescu 46 / 46
aspirations
vision
Summary
c© 2013 Andrei Alexandrescu 46 / 46
aspirations
vision
people
