HNHacker News
TopNewBestAskShowJobs

garenp

148 karma · joined November 2, 2011

submissionscomments
garenp··on Memoize – a replacement for make relying on strace
ClearCase has a clearaudit command that can be used to kick off any other arbitrary command and get dependency information (clearaudit /c <foo>). This is much better than using clearmake, because you otherwise depend on a users ability to (correctly) write makefiles upfront.
garenp··on Sergey Aleynikov Sues FBI Agents Who Arrested Him
I can totally believe that account. The people at Goldman were probably totally sincere in their belief that he had "stolen" their magic sauce when talking to the FBI.

These are presumably the same kind of people that think anything that's source code is some kind stuff that lays magical golden goose eggs, regardless of how trivial it is -- or even if it's open-source.

IMHO there is a much bigger, broader problem with non-technical people in positions of influence (and even the general public!). It's this same kind of disconnect that leads to mountains of trivial patents being filed. (I can't even count how many times I've heard of an engineer being put in a situation where they're told they need to patent some of their work, regardless of whether the engineer(s) think it's worth doing.)

garenp··on When Science Becomes News, The Facts Can Go Up In Smoke
Humor aside, it really concerns me that so much of the information in mainstream science news articles can mislead readers.

More worrying is that it's often not possible to persuade someone who is swayed in the wrong direction, because they just don't have the base level of knowledge to allow it.

On this point I almost want to say that every person who graduates from High School ought to have gone through a rigorous class in logic and another in statistics. It's all well and good to say everyone should have "critical thinking" skills, but you can't get there without some pretty solid intellectual tools.

garenp··on You never did math in high school
FWIW, I'd probably put myself in the same category the OP writes about.

I hit a wall in calculus 2 in University, because previously I had been able to rely on simply showing up and not putting in much effort to still do "pretty good". That was the first time I did poorly in any math/science course, and was a pretty big wake-up call. As a result, I ended up having to study more to fill in the gaps.

I was naive, came from a poor background with crappy public schools, and didn't think of math as particularly important to focus on until I saw how useful it was in physics. I don't think it's an all too uncommon story.

garenp··on Intel Clang-based C++ Compiler [pdf]
A given "feature" does not behave identically on different CPUs. To generate optimal code, you need to know which model is being targeted to make codegen decisions. It is simply not reasonable to expect Intel to do the work on their compiler to figure that all out and test it for competing vendor chips.

If you want more, use a different compiler, period. $VENDOR spends money on their compiler to support their chips, not to appease every whining, butt-hurt user on the Interwebz that thinks $BIGCO owes them something on a competing chip.

garenp··on Intel Clang-based C++ Compiler [pdf]
So they should just then believe what another third-party vendor CPU says, and ship the compiler and hope for the best? Not what I would do. I would only enable specific optimizations when I could guarantee the output would be optimal or a given model. "Hope is not a plan."
garenp··on Intel Clang-based C++ Compiler [pdf]
Not generating model-specific code for a processor that doesn't match your CPUID table is hardly what I'd call "sabotaging."
garenp··on Intel Clang-based C++ Compiler [pdf]
"Intel does not guarantee the availability, functionality, or effectiveness of any optimization on microprocessors not manufactured by Intel"

Right, because Intel should spend their finite resources buying every imaginable Intel-compatible CPU from other vendors just to test and qualify them for the Intel compiler.

garenp··on Burnout Comes in Three Varieties
Become a manager. :)
garenp··on PostgreSQL Columnar Store for Analytic Workloads
Probably referring to this:

http://blog.jooq.org/2013/08/24/mit-prof-michael-stonebraker...

garenp··on PostgreSQL: Jsonb has committed
Your spidey senses probably just need to first think "non-native English speaker?" :)
garenp··on Python 3.4.0 released
Which library is that? Truly curious.
garenp··on How Adobe Got Rid of Traditional Performance Reviews
It's striking to me that I've never seen or heard of a performance review process that allows for employees to provide feedback on how well their manager is doing. Sometimes a manager is just not a good fit for the group that they're managing, and like all people they too make mistakes, but all the efforts to "measure" things does not apply to them, which I can't imagine should exist in a universe that has the word "fair" in it's vocabulary.
garenp··on How Adobe Got Rid of Traditional Performance Reviews
The problem with setting very concrete goals like this, is that whenever you don't precisely meet them, they can be made to be a knock against you. As a result, it's then in your interest to set goals that are as easy to achieve as possible. I can't imagine that aiming for mediocrity serves the company or the employee well.
garenp··on How Adobe Got Rid of Traditional Performance Reviews
It's pretty notable, given that Adobe is a fairly large and well known company that is dropping performance reviews altogether.
garenp··on Simple Dynamic Strings library for C
Yeah, of course I'm aware, it's just that there is a time/space trade-off, so I think folks who argue that goto should never be used or has no legitimate uses, are just taking too much of a hard-line.

And I think it just highlights a lack of understanding about what might make goto "bad" in programming--IMHO it's "bad" when it makes control flow more complex, which undermines maintainability. Common "cleanup" idioms that only jump forward are not that bad, because they don't make control flow particularly more difficult to follow. (Whereas jumping backwards, especially across many state changes, can be very hard to follow.)

garenp··on Simple Dynamic Strings library for C
In performance critical nested loops, goto is the only way to exit early. A typical example here is performing a matrix multiply. In cases like this, it is absolutely necessary, because C doesn't provide a more granular 'break'.
garenp··on Scaling Mercurial at Facebook
Sometimes the cure is worse than the disease.
garenp··on Putin dissolves state news agency, tightens grip on Russia media
He was stuck in the transit area while waiting for a connecting flight, when the U.S. revoked his passport. Not much else to do but stay there.
garenp··on Visual Studio 2013
I'd like to see them automatically versioned and the "right" one selected at load time somehow. .NET can do it with assemblies, and it's a great idea.
garenp··on Why Russians Are Not Smiling (2011)
So as soon as you looked solemn/depressed on a regular basis, people started flocking to you? :)
garenp··on Hstore development for 9.4 release
800MB is for the old GIN index, the new experimental one is 350MB, and has not yet undergone more space optimizations. They are also not really comparable: Mongo only indexes a few select keys, whereas the new GIN index does many.
garenp··on Hstore development for 9.4 release
Look at some of the linked PDFs. It's the indexing technique and on-disk format that is an enabler for several notable use cases; personally, I'm looking forward to a time when I can use pg for full-text search without having to resort to another external system like Lucene or Sphinx. What it won't have is the query "smarts" like Lucene or Sphinx do today, but you need speed first.
garenp··on Mercurial 2.8 Released
If you're using NFS, CIFS or smbfs to host or synchronize repo data, You're Gonna Have A Bad Time (YGHABT?) regardless of which VCS you're using - it's only a matter of time.
garenp··on Mercurial 2.8 Released
Facebook tried to use git but ran into some scalability limits due to gits design (frequent index updates), whereas mercurial is append-only and doesn't have this problem. Hence, facebook has some legitimate technical reasons to not use git -- contrary to the common opinion that git is always "fast".
garenp··on Mercurial 2.8 Released
Totally agree with the first point in particular. The only reason I started using git is because it's required for github. Otherwise I'd use Mercurial exclusively.
garenp··on Source Code for the book “Machine Learning Systems with Python”
In my experience, the Packt books have always felt as though they have had very little editing and review (things you'd never see in a technical book from e.g. Addison Wesley or Prentice Hall). So if the book seems great it's probably due mostly to the author (which is fine with me if the book is cheap).
garenp··on Ask HN: What are the best technologies you have worked with in 2013?
PostgreSQL
garenp··on My first project in Rails – Pastebin for Mathematicians
Very nice. I always wished there was a nice way to write natural looking math via IRC--that was still readable in text form--with some kind of whiteboard looking interface for DCC chats that looked pretty.
garenp··on Optimizing C++ Code: Dead Code Elimination
If an empty loop does no computation (i.e. has no observable side-effects), then arguably there should be no difference in observable behavior from the program. Of course, sometimes we "know" something that the compiler can't "see", and so must indicate as such in our code (e.g. using something stored inside the loop, adding volatile, ...).

Probably the most common (or at least my favorite) example of DCE leading to a surprise, is when embedded programmers use so-called "delay loops" which can be eliminated by the compiler because it too is "empty" (the loop produces nothing that is used outside of the loop).

← PreviousPage 2 of 3Next →