The Whiteboard - 2013-01-28

« first day (878 days earlier) ← previous day next day → last day (4130 days later) »

user55340

3:33 AM

@JimmyHoffa I don't answer too many questions while at work... and don't have questions while not at work.

user55340

The electorate badge is one that takes a lot of voting on nearly every question (and avoiding voting on too many answers). The later part of that seems a bit counter goal for the site, though it doesn't mean I can't go back later.

12:32 PM

@NathanC.Tresch nice work on the updates to your question - I've even +1'd it for you now too. :D

1:24 PM

WTF we've got a (reasonably good) question migrated from SO but "source" question referred to by clearly visible link at "Programmers version", does not give reader a clue about that. All one can see there is closed-off-topic...

56

Q: Is Ken Thompson's compiler hack still a threat? (yes/no/why?)

Ken Thompson Hack (1984) Ken Thompson outlined a method for corrupting a compiler binary (and other compiled software, like a login script on a *nix system) in 1984. I was curious to know if modern compilation has addressed this security flaw or not. Short description: Re-write compiler code t...

linux unix compiler hacking

notice says migrated from stackoverflow.com 2 days ago that's a good way to the source, great - but why source question doesn't have a similar "migrated to" note?

this feels like pretty klingonian way of doing things...

user image

2:19 PM

@gnat I'm afraid that's a side-effect of the migration rejection mechanism. The question was closed after it was first migrated, which removed the migration notice from the SO version. Then, it was re-opened.

3:10 PM

@YannisRizos thanks, now that makes better sense (it would be nice for migration notice to automagically recover upon reopening, but that's more of a corner case)

@gnat It might be worth it to post a bug report on MSO

neener neener my answer is better than FrustratedWithFormsDesigner's, even if he does have wayyy more points than me, and will likely get more votes because he got his answer in first

Translation: I'm a winner in my own mind!

3:54 PM

@YannisRizos it's already there: meta.stackoverflow.com/questions/155249/…

user55340

4

Q: What is a good script language for writing server administration tasks?

I am administering hundreds of RHEL servers. There are many daily tasks to perform. Right now I am working with bash scripting and python. I was wondering if other languages such as Erlang, Lua, Ruby, Haskell, or Clojure.

multithreading concurrency scripting

user55340

Think of doing sysadmin tasks in Haskell...

4:05 PM

@MichaelT I suggest it!

:)

@MichaelT What makes scripting languages helpful? You can do a lot ad hoc with very little code. I don't know of a language you can do more with less than haskell

it may be garbage and panned if you ever show it to any of the category theorist nuts out there, but so are sysadmin perl scripts shown to some professional perlista I'm sure.

Damnit I just tried finding the document I read a few months back about how google started using Haskell for sys admin tasks rather than Python, and someone wrote a huge experience report on it, but now it's paywalled!

Ah here it is!

k1024.org/~iusty/papers/icfp10-haskell-reagent.pdf

ganeti-htools are a bunch of sys-admin tools they wrote in haskell and released open source and that document is about the sys admin team that made them and their experience moving to haskell from python

4:26 PM

@MichaelT I was actually talking just yesterday with some folks about whether it makes more sense to use a parsec over regexp frequently and when, the general consensus was it depends on the use but often times it does. Not long ago I wrote in ~15 minutes a snippet of haskell that removed all styles from all files given a filename filter (like *.aspx) in a directory recursively. Doing the same with regexp would have taken me at least as much time, though that may be an affect of my unfamiliarity

rather, it removed arbitrary attributes from an entire xml document, both with and without quotes, you just feed it a list of the attributes

probably not a difficult regexp though now that I think about it.. something like attribute=["?][.*][" ] or some crap

(I really don't know regexp, what a crappy dev I am)

user55340

A regex is just a representation of an NDFA. If you can understand NDFAs, you know regex.

user55340

And while I do mock haskell for sysadmin, in that future where perl and haskell run on the same virtual machine, haskell will access to all of perl's libs. And from what I understand, perl 6 is being influenced by haskell.

user55340

All that said, I'm fairly sure that "professional perl types" (which I was one once) are aware of the "get it done" world of sysadmin perl (which I was one once) is quite valid. Writing beautiful libraries all day means nothing if they aren't used in the real world.

4:42 PM

@MichaelT I'm just saying, the "get it done" approach can be used by haskell to great effect, the google dude who worked on those sys-admin tools in haskell specifically said after a year of working in haskell on that stuff he really still didn't understand or know just about any of the higher level stuff in haskell that may have made his code simpler or better, but lacking that knowledge he had no troubles forging out some greatly helpful tools

user55340

The biggest argument for perl (or python) for sysadmins is that other sysadmins are likely to know it. The "get it done" aspect also entails "some other sysadmin can fix it quickly." Having "haskell" as a job requirement for a sysadmin could rule out quite a few candidates.

the code would have the same differences from Galois Quants that perl does between get-it-done sysadmin to professional perler which you are familiar with. That said, it would be quick and effective. That is interesting that perl 6 is trying to steal ideas from haskell; that seems to be what ever language is doing heh. Python took comprehension syntax from haskell, C# took linq from haskell

Oh yeah, no doubt, that was the biggest roadblock the dude noted in his experience report:

4.5 High barrier to entry
Lastly, we believe that the most signiﬁcant problem is the high
barrier to entry.
Even after the completion of this project, the author feels that
his knowledge of Haskell is very much incomplete, and that he
is far from being familiar with advanced topics (e.g. applicative
programming, generic programming, etc.). Whether this is needed
or not for small projects is debatable—for example, our current
code works with only standard Haskell 98 (no language extensions
in use)—but it might be possible that careful use of advanced

user55340

Over the years, Perl 6 has undergone several alterations in its direction. The introduction of concepts from Python and Ruby were early influences,but as the Pugs interpreter was written in the Haskell programming language, many functional programming influences were absorbed by the Perl 6 design team.

user55340

(From wikipedia)

and as for understanding regexp, there's a difference between understanding a state machine as you talked about and actually knowing the syntax to regexp, which I can never happen to memorize since I use it so infrequently

haha that's funny. Not surprising, writing parsers in haskell is so effing easy

(which is why I opted for it over regexp for my small automation task)

and why a DSL tends to be the solution approach to most problems that haskellers take, makes sense, imagine if in your favorite language it were unbelievably easy to write parsers. Makes for new opportunities, because in most languages, parsers are freeging pains in the arse.

thinking about it, over the years I'm pretty sure i've written considerably more xpath than regexp, and if that's not true I tend to remember xpath better... I think because xpath uses rational symbolography (made up words are fun!) to the underlying language, I don't recognize regexp symbols in a meaningful way with relation to the underlying language (other than the simple string match)

@MichaelT that's funny, pugs is hosted on hackage

user55340

4:56 PM

The symbols of xpath are about as confusing to a newcomer as the symbols of basic regex. All you really need to know to handle most problems are () for grouping, * for 0 or more, + for 1 or more and [] for character classes.

ah, yeah I never remember that. What's the . mean?

user55340

Any character.

and ? is 0 or 1 of any character?

user55340

Yep.

I just don't use it enough. I can never remember the description you just gave, I should actually write that down because that is a simple explanation heh

user55340

4:58 PM

(gah, meeting time)

6:09 PM

Was that DB question migrated to DBA?

Balls, I almost suggested that but I'm afraid of it getting closed as dupe on DBA, even though the question it's a dupe of has a frankly garbage answer.

6:19 PM

flagged for moderator attention over there that maybe they'll unclose it, unlikely but wth. The 3 answers on it right now are vastly better than the answers on the question it's claimed dupe of.

user55340

Write a better answer in the original one? Or maybe merge the two and bring the other answers into the original one?

Kaushil Rambhia

guys i m 3 months experience in php n html css but i want to improve my programming logic...what can i do for it

Write more code

become 6 months in

then 6 years in

then 10 years in

Kaushil Rambhia

dude u told same thing which i asked

i wan to improve my logic not writing syntax..

user55340

What part of the 'logic' are you trying to improve?

Kaushil Rambhia

6:35 PM

like how n wer to place loops,build sql quries,

user55340

Write something that needs to use those structures and do it.

Kaushil Rambhia

how to improve that kind of thinking...so i can write even a small code but a perfect code

user55340

I would tend to suggest a language that isn't as... tossed together as php and work on some more 'pure' programming problems.

Kaushil Rambhia

currenlty it takes me 10 edits to make a perfect code..

should i try c ?

for improving logic?

user55340

The language isn't as much of a problem. Pick one and work with it.

user55340

6:39 PM

I would tend to suggest trying to do Project Euler problems. You will need to think about how to do it to work these out. Only worry about the first 50 (the more complex ones are... more complex and into deeper math problems).

user55340

Consider the first question - find the "sum of all multiples of 3 or 5 below 1000"

36

Q: I can write code...but can't design well. Any suggestions?

I feel that I am good at writing code in bits and pieces, but my designs really suck. The question is, how do I improve my designs - and in turn become a better designer? I think schools and colleges do a good job of teaching people how to become good at mathematical problem solving, but lets adm...

learning career-development professional-development developing-skills

40

Q: How do you go from a so so programmer to a great one?

How do you go from being an okay programmer to being able to write maintainable clean code? For example David Hansson was writing Basecamp when in the process he created Rails as part of writing Basecamp in a clean/maintainable way. But how do you know when there is value in a side project like...

design career-development maintenance

23

Q: What are your suggestions on learning how to think?

First of all, this is not the generic 'make me a better programmer' question, even though the outcome of asking this question might seem similar to it. On programmers.SE, I've read and seen these get closed here, here, here, here, and here. We all know there are a multitude of generic suggestion...

learning thinking self-improvement

@YannisRizos I am baffled none of those are closed

@YannisRizos can I close vote them all as off-topic?

@JimmyHoffa Look closer.

Oh, one of them is closed

weird, it didn't show [Closed] in the topic

Kaushil Rambhia

6:43 PM

@MichaelT thanx for euler

32

Q: How can I improve my problem-solving ability?

Everyone says the same thing: "a real programmer knows how to handle real problems." But they forget how they learned this ability or where: it's not taught in schools. What can I do to improve my ability to tackle complex programming problems? What strategies have worked for you? Are there spec...

self-improvement problem-solving

@KaushilRambhia I would advise design principles if you were further along, but those are for designing where you're still working out how to think logically about this stuff. I genuinely believe the thinking logically to come up with implementable solutions to problems literally comes from practice and nothing else. However, I was being facetious earlier, that initial logic part takes maybe ~a year

after your first year in industry, in a real job writing code (html doesn't count), fixing bugs etc you will have built the practical logical solution part of your brain to know how to think in control-flow, unless you start with FPL then you'll think in contexts and transformations instead, but most go imperative

user55340

Personally, my most fluent programing language is perl - thats what I tackle the problems with first. I'm learning clojure... and doing the problems again in a new language to try to learn how to think in the language.

also don't change languages a bunch for the first 6 months - a year

user55340

As for a language to do those problems in, the php similarties with perl would likely mean that perl would be a good choice to work with (that will directly translate back to php). That isn't saying don't do it in C or python (or if you want to melt your brain, haskell or clojure).

6:49 PM

@MichaelT for all my love for haskell, I still think in C#, I started in C as a lad, then VB, so that whole structure is still how my brain processes things. The other stuff is slowly seeping in though

@JimmyHoffa Be happy you started out with C. I started out with (classic) BASIC, I'm still not convinced GOTO is a bad idea.

user55340

Working with a functional language does melt your brain, but that makes it rather fluid and plastic (in the physics sense of the word) and keeps it from getting brittle.

@YannisRizos it's not a bad idea if you like to have no idea how your code actually works

user55340

@YannisRizos ever read Story of Mel?

Kaushil Rambhia

thanx guys for support..what i wanted was to train my brain to generate logics..because today i m wrkng in php next time might be in android...so after learning new syntax i should have capabilities to write efficient n in less time

user55340

6:51 PM

catb.org/jargon/html/story-of-mel.html

user55340

The new computer had a one-plus-one
addressing scheme,
in which each machine instruction,
in addition to the operation code
and the address of the needed operand,
had a second address that indicated where, on the revolving drum,
the next instruction was located.

In modern parlance,
every single instruction was followed by a GO TO!
Put that in Pascal's pipe and smoke it.

@MichaelT I would say if you didn't know imperative first, I doubt FP has that same meind-bending effect. You spend years working your brain up to think in control-flow as a way of making things happen then follow that up with trying to convince your brain you can make stuff happen simply by describing what it is you want rather than telling it how to happen?

user55340

Datamation published a way to have goto-less programing with the come from statement.

user55340

fortran.com/come_from.html

that mind-bending effect is the effect of everything you spend years learning being destroyed in front of your eyes

Kaushil Rambhia

6:56 PM

amazon.com/dp/0141003146/?tag=stackoverfl08-20 anyone know anything about this?

I vaguely remember the "If a program can't rewrite its own code,
he asked, what good is it?" line, so I probably read the whole story at some point.

@KaushilRambhia Haven't read the book, but it doesn't seem to be focused in programming logic...

@KaushilRambhia a lot of times "logic" type stuff is going to be more mathematically based logician stuff which will be in line with FP but not the type of programming you will likely be doing your entire career

user55340

@KaushilRambhia That is dealing with logic as a philosophical discipline. Determine if: (a ^ b) -> (c v d) is true when a is true and c is false.

@KaushilRambhia there are "logic" languages like prolog that work exactly based off that stuff, but they are nothing like the languages you will use in industry unless you get really lucky

Kaushil Rambhia

hmmm

7:00 PM

What michael just described is stuff you would see in those languages, but like I said, if you can't tell; what he just wrote is nothing like php or java or perl etc.

Blog - Stack Exchange: Podcast #41 – Neither of Us Have Muscles

Alex Miller on January 28, 2013

Welcome to Stack Exchange Podcast #41, featuring Joel Spolsky, Jay Hanlon, David Fullerton, Kyle Brandt, Nick Craver, and Geoff Dalgas, with Producer Alex calling in from Denver! We have a bunch of systems administrators and the like here, because we are in the process of moving datacenters to our new home in New York City.

So what’s involved in the move? We hired movers to do all the de-racking and truck driving, so the work done by SE employees involved laying everything out and then wiring it back up. …

(I can't imagine even perl 6 goes as far as to allow for concepts even remotely close to that heh)

user55340

I had a class from the philos department back when I was in college on logic. When you get to the updside down A and backwards E symbols, it gets real fun. Incidentally, CS majors were the second largest group in the class after Philos majors.

user55340

You get to things like this...

user55340

everything2.com/title/Tree+method+for+logical+decomposition

7:01 PM

@MichaelT yeah, learning haskell I've learned what those symbols mean because they're all over haskellers documentation. As are logical/mathematical laws like definition of an inverse and all kinds of shit

user55340

1) A and B x
----------
2) A
3) B

user55340

Where the statement "A and B" becomes two seperate statements of "A" and "B" in the tree.

Was funny, trying to teach a colleague some FP recently I had to make him write out a truth table for something he was doing because he just couldn't figure out what inputs/outputs he was going to get

Makes sense why logicians actually use those things

user55340

Need to work on that... the formatting was just wrong there.

@MichaelT Yeah, I've tried stuff like that before. Spaces are optimized heh

Kaushil Rambhia

7:04 PM

bye guys.thanx for help gn

user55340

Anyways... that's logic.

user55340

Tangentially, in AI class back in college, there was an assignment to write a logical decomosition tool in lisp. Given a statement, identify what values for the variables are necessary to make it true.

@MichaelT So you wrote LISP in college but are struggling with Clojure now?

user55340

I had a function named "hamletP" that identified expressions of the form '(or 2B (not 2B))' more quickly.

@JimmyHoffa I wrote LISP in college and I'm struggling with... LISP now.

user55340

7:06 PM

It was one class, and as one lisper in the lab I hung out said "I wrote beautiful C in lisp"

haha fair enough

I guess that goes along with the reason neither of you could derive a differential equation now (I think I used those terms correctly, never took calculus)

user55340

The practical functional things such as "infinite sequences that are lazily evaluated" were not covered. Functions that work on entire lists where glassed over.

eh, that's practical to an extent. Lexical closures are just as practical I would say

which is what LISP is all about after all

user55340

And remember that college was nearly two decades ago.

aye

user55340

7:11 PM

Lets see... the assignments I had back in AI in lisp... there were three of them. One was that logic one. One was a best-first scheduler. I forget the other. The last (not lisp) problem in the class was a simplified Go-Maku (5 in a row on a 19x19 board).

@MichaelT I would be curious to see what you're hacking up in clojure if it's not for company, you should github it. I been pondering learning clojure myself, I quite like the nesting style of lisp which seems to turn people off, I feel it makes scope very explicit

user55340

I did it in C++ (I think it was ++). I translated the 19x19 board into a 192x5 board and had it "play" that one.

user55340

My clojure hacking is for just "know it"... keep the brain molten. I attempt to do Euler problems with it.

user55340

(and those, as a nice Euler participant, I don't publish the code for).

ah

most people publish their eulers I thought

user20683

7:14 PM

@JimmyHoffa I don't

interesting

user20683

I'll probably put them up on a locked site for interviews

user20683

since they are a mildly decent demonstration of applied mathematical ability.

I figured it was a leave it up to the practitioners not to read my code, because at the end of the day every euler problem has posted answers in most every language

user55340

If you look, you can find the posted answers in just "here's the answer"

7:16 PM

If you look you can find posted code for all the eulers you want

google.com/…

You're not putting anything out there others haven't already put out there

it's up to each participant not to look up the code and figure it out themselves, I figure

user20683

@JimmyHoffa I need to get around to writing another blog post

user20683

kickstart that thing again

What thing? heh

Oh the blog?

user20683

@JimmyHoffa yeah

What are you going to write it on?

user20683

7:18 PM

@JimmyHoffa probably do a grab-bag of answers to stupidly common questions

user20683

similar to my "no silver bullets" post

I hate the no silver bullets concept

I agree people shouldn't buy into silver bullets, and people shouldn't believe in silver bullets

but people so often use no silver bullets as an excuse not to strive to create silver bullets

which is what we should all be doing at all times. You want a silver bullet? Pure true honest to god AI, like the real thing, consuming the internet on fiber.

user20683

@JimmyHoffa I basically argue that DSLs are good but if you are going to learn a language, you shouldn't be learning it as a silver bullet unless it's a DSL, that there is no "one true language"

user55340

There are many people trying to create silver bullets. The problem is, they are only silver plated lead, and they cost you extra.

Just because it doesn't exist doesn't mean we shouldn't try to creat it, is my approach to silver bullets

user20683

7:20 PM

my approach is very much the idea of DSLs

@MichaelT like I said, don't buy or believe in silver bullets, but we should be trying to create the perfect solution to each and every one of our problems

user20683

R is an excellent example of a silver bullet for Data Analysis/Display

user55340

Remember who The Mythical Man Month is for. Its not for the programer... its for the manager that keeps wanting to find that silver bullet.

we may never create perfect solutions, but at least in trying we don't limit the quality of our solutions

user20683

@MichaelT The silver bullet is easy, listen to your programmers before you go meet with your clients ;)

7:22 PM

@MichaelT I recognize that, but I'm just tired of hearing "There's no silver bullet" parrotted by engineers to eachother to excuse the crappy quality of their code

user55340

That doesn't make it an order of magnitude faster, it makes it not an order of magnitude slower.

user20683

@JimmyHoffa code quality is a different issue from what I discussed.

that's my problem with "no silver bullet", it's an excuse for defeatism and it comes from a credible source so it carries weight

@WorldEngineer I understand that, but I think "no silver bullet" is a catch phrase that shouldn't be spread even if there are rational interpretations. There are good uses of goto, but you never tell someone they should ever use goto, and similarly you should never tell some one there's no good solution, because they'll say great, I'll create a shitty one, with gotos.

I just don't like the catch phrase and think people should stop spreading it.

it has more negative impact than positive because the people it's meant for are those ignoble PM's who won't even listen anyway

2 hours later…

8:56 PM

@WorldEngineer I still need to learn R

On the other hand, I've an interview Wednesday and genuinely fear doing something too functionally styled from recently built habits and simply confusing the interviewers. Need to go practice data modelling to think in C# style OOP or something.

9:17 PM

This is pretty cool:

stack-ed.com

busted CSS

make your browser skinny and it moves the questions down to the bottom

(I always use vertical monitors, so my browser's considerably skinnier than most)

nice prettify on the answers though, preferable to SO's actually I think'

« first day (878 days earlier) ← previous day next day → last day (4130 days later) »

Transcript for

Jan '1328

The Whiteboard

General Discussion for softwareengineering.stackexchange.com P...

coffee-day horror-stories monads my-code-is-compiling scotch the-other-monitor

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

site design / logo © 2024 Stack Exchange Inc; legal

mobile