« first day (1802 days earlier)      last day (3192 days later) » 
04:00 - 18:0018:00 - 00:00

Nick Alexeev
Nick Alexeev
4:35 AM
@gnat I don't see this as broken windows. I see it as heads on spikes.
Programmers.SE is a tough stack. (My kind of stack.)
 
Duga
Duga
5:07 AM
You might look at the topic guidelines / site rules for programmers.stackexchange.com to see if this would be more welcome there. — Charles Duffy 14 secs ago
 
gnat
gnat
5:54 AM
4
A: Can we track the positive effects of a popular question?

jmort253Ever heard the phrase "a flash in the pan"? According to my mom, my grandfather and grandmother allegedly used this phrase to talk about Elvis Presley when he made his first break. The idea was that he would have a brief "flash" of popularity and then evaporate into nothingness; he wouldn't even...

 
Nick Alexeev
Nick Alexeev
6:14 AM
@gnat Pardon my lack of understanding, but I don't see how that thread on meta.Workplace.SE relates to broken windows (or heads on spikes). What are you trying to say, @gnat?
 
Robert Harvey
Robert Harvey
@NickAlexeev Apparently you've never experience Math Overflow.
They were so hardcore that they broke free of the SE network, because they thought we were pansies.
 
gnat
gnat
@NickAlexeev I am trying to say that these spikes are well known to be useless at best. As for how broken windows "work", I provided concrete recent example in comments over there
today's example for those interested how broken windows... "work": 10K link / screen shot. Asker didn't pay any attention that "inspirational" question has been closed (FWIW it wasn't even recent nor high score) — gnat 16 hours ago
 
Nick Alexeev
Nick Alexeev
@gnat Oh, when I mentioned heads on spikes. I meant different spikes.
... not the kinds of spikes you see on an oscilloscope.
 
Robert Harvey
Robert Harvey
There are a lot of questions on the front page that just need to go away.
 
gnat
gnat
6:31 AM
@NickAlexeev understood now. :) The problem is, these heads stink. And make more stinky stuff get here. As Atwood pointed in that article I refer, "these posts... truly start to drown out everything else on the site..." For another interesting example, you can search MSO or MSE when Shog blacklisted code-golf at SO, letting it just stay closed turned out not enough
 
Nick Alexeev
Nick Alexeev
@gnat I have read your post on meta more carefully and some of the referenced works. I agree with you that the Hot Network Questions feature often promotes wrong kinds of questions.
I see a similar thing on EE.SE. My eyeball estimate is that 30% to 50% of EE.SE questions that got hotted, are ones that I would not promote.
 
gnat
gnat
however simple it is, I don't expect this (or any other solution for reported issue) to be implemented in foreseeable future, because its impact on Stack Overflow seems to be negligible. Substantial, direct benefit of proposed feature would likely be achieved only on smaller sites — gnat Feb 2 at 22:02
...another recent example: Where would one choose C11 instead of C++?. Of course I can't read asker's mind and be completely certain that it's inspired (possibly unconsciously) by Is it better to store data in a small, unreadable format, or a large, readable format? but I can perfectly imagine asker wondering why their question is getting downvotes when just a week old, similar Gorilla vs Shark sits comfortably at +28. — gnat 1 min ago
 
Duga
Duga
7:34 AM
StackOverflow is for code questions, something when you're stuck and you tried so many other things... if your question is regarding an idea, you should post it to Programmers — balexandre 12 secs ago
 
 
2 hours later…
Duga
Duga
9:34 AM
I think you should ask your question here — Zack 32 secs ago
 
ratchet freak
ratchet freak
10:02 AM
@Duga heh a user that got referred to progs.SE is critical about the advice
 
 
2 hours later…
Thomas Owens
Thomas Owens
11:54 AM
Fantastic. There's a widespread Outlook / email outage across the corporation, affecting everyone.
 
ratchet freak
ratchet freak
@ThomasOwens license issue probably
or server crash
 
Thomas Owens
Thomas Owens
No idea, but there's no ETA.
That also means Lync is down. So no IM, no email.
 
gnat
gnat
@ThomasOwens if you have ideas on how to improve IT infrastructure, this is a good time to push these. Such outages sometimes (unfortunately not always) make management realize that what they considered spending is actually an investment
 
Thomas Owens
Thomas Owens
@gnat Except our IT infrastructure is managed by contractors.
We seriously outsources everything. 100%. Total infrastructure. We control nothing.
The extent of the IT staff employed by the company exists to manage the contractors for different things - workstations, servers and storage, telephones, printers and faxes, mobile devices.
 
ratchet freak
ratchet freak
tell them to outsource management as well
2
 
gnat
gnat
12:08 PM
@ThomasOwens here's your idea for improvement. Source it back home
...your company won't be the first who went that route
 
Thomas Owens
Thomas Owens
We are bringing in an internally managed engineering development network because of it. But things like email are managed like 5 levels higher than my manager.
So we'll be running our own servers for development, version control, building.
 
ratchet freak
ratchet freak
@ThomasOwens run a small irc server for emergencies
especially if they also removed telephones in favor of lync
 
gnat
gnat
@ThomasOwens since all of these levels are affected by email / IM outage, it will be easier than usual to pass it up the chain. You only need to account for that higher level lingo to make it go up smoother...
72
A: How can I tell people to get to the point?

KazYou need to forge a new mission statement which establishes brevity as a core value for your organization. Then hold meetings to find stakeholders who will buy in to the innovative paradigm. Be sure to mark down action items for everyone, and track them so that they follow through with their new ...

 
Thomas Owens
Thomas Owens
We still have telephones. But I should throw out the idea of IRC. I'd also like to throw up one of the open source Stack Exchange clones, unless we include some kind of JIRA plugin that handles knowledge. Although the project SharePoint repository is the "ultimate" thing.
@gnat Believe me - people are scrambling to fix the email outage. I called the help desk number and they provided ticket numbers to track. People have been working it for like 4 hours now.
So I'm putting together a question for Prog. I'll link it when I'm done if anyone wants to answer it.
 
gnat
gnat
@ThomasOwens mission statement is to source it back home from offshore... as a core value for your organization :)
"stakeholders will buy in to the innovative paradigm"
 
Thomas Owens
Thomas Owens
12:18 PM
It's not offshore. It can't be offshore because of the work we do. If it was offshore, maybe they would have a cost savings. But instead of paying salaries + benefits, they are paying an third party to pay their people salaries + benefits.
 
gnat
gnat
@ThomasOwens it's a known thing. Offshore or not, outsourcing to 3rd party is already proven as not always good idea as it was thought before. "Innovative paradigm" is frankly not very innovative. Many companies learned the hard way to count their losses from outsourcing and figure that some things are just better to bring back home
 
Thomas Owens
Thomas Owens
Everyone here agrees, and as far as can tell, that goes all the way to the GM.
I wonder how I know when this email outage is done. I closed Outlook because it couldn't connect to the server and kept prompting me for a password...eventually it will be fixed.
 
ratchet freak
ratchet freak
@ThomasOwens when lync can log-in again
 
Thomas Owens
Thomas Owens
I closed that too for the same reason.
 
gnat
gnat
@ThomasOwens while I have your attention - Community user recently "outsourced" featured tag from that STCI Phase 2 question at meta, could you please bring it back home? I guess this tag auto-expires after a month
 
Thomas Owens
Thomas Owens
12:32 PM
Will do. Oh. There was another question about featured, but I tried to restart my computer to fix Outlook and lost the tab.
Harvey Balls
Harvey Balls are round ideograms used for visual communication of qualitative information. They are commonly used in comparison tables to indicate the degree to which a particular item meets a particular criterion. For example, in a comparison of products, information such as price or weight can be conveyed numerically, and binary information such as the existence or lack of a feature can be conveyed with a check mark; however, information such as "quality" or "safety" or "taste" is often difficult to summarize in a manner allowing easy comparison – thus, Harvey Balls are used. In addition to...
I was unaware of a name for these until just now.
 
gnat
gnat
Thomas, would you consider tagging this featured: discussion of recent popular questions. It may probably benefit of a wider exposire to site community — gnat 2 hours ago
^^^ this one?
 
Thomas Owens
Thomas Owens
Ah. Yes.
I thought it was a flag for some reason. It was a comment. Didn't think to look my in inbox.
I remember thinking how good the idea behind those questions were, but how terrible of a fit they are for SE...
 
gnat
gnat
 
Thomas Owens
Thomas Owens
@gnat That's not the most appropriate fit.
Unlike questions like farting in a cubicle or what programming name to give my pet or who your favorite computer scientist is or who you should follow on Twitter, these ones are about real-world work problems.
The first batch doesn't require professional software developers and isn't about solving a software development problem. The second set is.
 
gnat
gnat
@ThomasOwens per my reading, that general reference covers these as well, as part of a broader bucket. Q&A model turned out surprisingly robust, if you consider all... permutations it survived at different SE sites. Skeptics, Code Colf, Code Review, Software Recs and others bend it in many various ways. The only thing it couldn't handle so far (and probably wouldn't handle ever) is open ended discussion
 
Thomas Owens
Thomas Owens
12:43 PM
@gnat Open ended discussion? No, it doesn't.
I think the biggest barrier is people. If you could ask a question and people would post self-contained answers (after checking to see if another post was already the same answer), it would work, I think.
I do think there's a difference between a discussion and a question with a large amount of answers. The system isn't design for the first, and people are a barrier to effectively doing the second since they are more used to forums.
 
gnat
gnat
@ThomasOwens ah that. I am not sure, because no one seriously tried that, but I tend to agree. Having problem with that could very well be "implementation problem", not a model limitation really. That "MeFi pattern" of proactive protection, mentioned by Shirky, it suggests that there may be the way out. When a user with positive rep posts garbage, and it gets downvoted, and these downvotes aren't countered by lemming upvotes, one could only imagine how things could work then
no one tried that at SE
 
Thomas Owens
Thomas Owens
Programmers would be an ideal place to try something like that. But as an established site, I'm not sure that it wouldn't be confusing.
Many questions that we have when conceptually working on software don't have right or wrong answers. Just a range of answers that are more right or more wrong for our particular circumstances.
But writing a question in a way like that is...not easy.
 
gnat
gnat
@ThomasOwens that's an interesting twist. You may be on to something, I didn't think of it that way so far. One counter-argument I can think of from SE higher ups is that things already work more or less that way, in regular questions that aren't exposed to hot list lemmings. They could tell that going this way will simply break entertainment we provide to wider audience. "Give few questions to us to entertain the crowd, and enjoy the majority of the rest." This is possibly how they feel....
...as for broken windows effect, they likely don't care as it doesn't impact SO
 
Thomas Owens
Thomas Owens
Do you think that there are other sites that may experience the ability to have large numbers of answers to questions? I could ask their mods what they think, too.
The Workplace may (@endlerland).
 
gnat
gnat
1:02 PM
@ThomasOwens besides Workplace (which not coincidentally has a unique feature of immediate protection by 15Kers), unlikely. We're talking a site that has topics where typical SO lemming easily can and want to talk about (and vote up with their assoc bonus), it's not kind of stuff discussed eg at Physics. Also, there should be no "hard" posting rules like ones that protect Soft Recs / Code Review / Code Golf
 
Thomas Owens
Thomas Owens
Software Recs could be this way, but they require a specific enough set of requirements that it isn't likely to be more than a handful of answers.
I do want to think about this some more, though.
 
gnat
gnat
@ThomasOwens when the audience of potential answerers is sensibly limited, or when posting rules strongly impose certain specific effort to answer (and allow quick objective deletion if it isn't met), questions that naturally can have many answers, could in theory fly normally. At least until these bump into UX limitations - how many reasonably long answers can one grok in a single question?
but that's not like what we have at Programmers :)
 
Thomas Owens
Thomas Owens
Yeah. If I saw a question with 10 answers, I wouldn't want to read all 10 before posting my thoughts, I guess.
It would take me longer to read all 10 (assuming they were thought out and full answers) than it would to write my 11th well thought out answer.
 
gnat
gnat
1:18 PM
@ThomasOwens oh. This reminded me heretic idea I once posted at TWP meta. This way, questions with multiple answers would require progressively more effort to add a new answer. But this feature is unlikely to ever be implemented, I am really surprised that it gained so little downvotes...
8
Q: When there are many answers already, help me check that mine won't repeat others

gnatWhen posting my answer I check that it doesn't repeat others as required by our FAQ. In current UI, this becomes difficult when question already has many answers. I would want that questions with 10 or more answers (including deleted ones) have additional UI allowing me to review each of previo...

feature-request answers
 
Thomas Owens
Thomas Owens
I blame pagination.
Also, do you know how there's "suggested questions" when you're typing the body of a question?
Should there be "suggested answers" when you're typing the body of an answer?
 
gnat
gnat
@ThomasOwens yes, I even recall testing that "suggested questions" feature. It's inobtrusive, which is both strength and weakness. If I wanted to thoroughly check other's answers, I would much prefer comparing them side by side against a preview of what I drafted. This is what I try to do myself now, which turns out hard without system support when there are 5 or more answers already. As a result I end up generally abstaining of answering these - too hard to check
 
Thomas Owens
Thomas Owens
1:34 PM
0
Q: What method should I use to populate a tree as I'm consuming data?

Thomas OwensI have a tree that consists of 5 types of nodes - V, W, X, Y, and Z. V represents the root node, and a given tree has exactly one element of type V. A V contains one or more Ws, a W contains one or more Xs, an X contains one of more Ys, and a Y contains one or more Zs: Looking at this, I decid...

data-structures data trees
If anyone wants to answer.
 
user55340
@ThomasOwens 44 seconds. Still trying to beat oded's record?
 
Thomas Owens
Thomas Owens
@MichaelT I think he has a sub-20.
 
user55340
Did you see "the button" this spring on Reddit?
 
user55340
The Button (Reddit)
The Button was the name of a meta-game and a social experiment regarding an online button and countdown timer that would reset each time the button was pressed. The experiment was started on the social networking website Reddit on April 1 and was active until June 5, 2015. The game was started by a Reddit administrator and was thought to be an April Fools' joke. The Button garnered enthusiasm from Reddit users worldwide, attracting clicks from over one million unique user accounts. Various websites, browser extensions, and mobile applications were created for the sole purpose of tracking the live...
 
Thomas Owens
Thomas Owens
@MichaelT Yes. I'm a non-clicker.
 
user55340
1:39 PM
Consider the ones who got 60s
 
enderland
enderland
1:55 PM
> Responsible Mgr: NULL TEST
ok then
 
Jimmy Hoffa
Jimmy Hoffa
Happy coffee day
 
durron597
durron597
2:57 PM
@MichaelT Got my first three minute closure today. Not bad for a baby mjollnir
So, in summary: We should create a canonical Q&A thread where the canonical answer is simply "Install a different IDE, and this problem goes away". — Yakk 22 mins ago
 
Robert Harvey
Robert Harvey
3:34 PM
@durron597 What's the problem there?
 
ratchet freak
ratchet freak
@RobertHarvey too chatty
 
durron597
durron597
@RobertHarvey Lots of too chatty, and comments that should be edited into the answer
Including your own like:
Is this: *Background Color <%=Html.RadioButton('rbBackgroundColorWhite', 0, false)%>White <%=Html.RadioButton('rbBackgroundColorWhite', 0)%>Black in the browser or in VS2008? — Robert Harvey ♦ Jun 17 '09 at 3:26
 
Robert Harvey
Robert Harvey
That was just a clue.
 
durron597
durron597
I guess I should have just flagged, I wanted to give you the opportunity to save some of the content there with edits
 
Robert Harvey
Robert Harvey
Don't get too chapped with the comments. Post-it notes can theoretically live on a monitor indefinitely, y'now.
Assuming they are useful. I deleted some of them.
 
Thomas Owens
Thomas Owens
3:40 PM
I wish that Java ArrayList's add(int, E)worked by adding null elements to fill in the gap between the end of the list and the newly added element.
 
durron597
durron597
@RobertHarvey Oh wait, I just realized you have two answers there
Should one of them be deleted?
 
Robert Harvey
Robert Harvey
[sigh]
 
durron597
durron597
okay fine you don't care I understand
 
Robert Harvey
Robert Harvey
Well, conceptually, they say two completely different things.
I don't generally post two different answers unless they really are two different answers.
 
durron597
durron597
Yeah but that was from 2009
before your razor sharp wit was fully honed
 
Thomas Owens
Thomas Owens
3:43 PM
I SOLVE IT!
 
ratchet freak
ratchet freak
@ThomasOwens you can make a List that does that
 
Robert Harvey
Robert Harvey
\o/
 
Thomas Owens
Thomas Owens
@ratchetfreak Yeah. Or I can use ensureSize(). It's two method calls. But I just solved my problem. A little prompting toward an adjancency list helped, but it's also more robust and just a better data model.
 
ratchet freak
ratchet freak
there is no ensureSize in java.util.ArrayList only ensureCapacity which works like std::vector::reserve
as in it doesn't affect the size only the backing array
 
Thomas Owens
Thomas Owens
@ratchetfreak Yes, but it seems like that will add the indices that I need. So if I have a backing array of size 5 and call ensureCapacity(10), I now have indices 0 through 9 available.
That's what the documentation says.
 
ratchet freak
ratchet freak
3:52 PM
you are subclassing ArrayList?
 
Thomas Owens
Thomas Owens
No, why would I subclass it?
 
ratchet freak
ratchet freak
capacity is distinct from size
> Each ArrayList instance has a capacity. The capacity is the size of the array used to store the elements in the list. It is always at least as large as the list size. As elements are added to an ArrayList, its capacity grows automatically. The details of the growth policy are not specified beyond the fact that adding an element has constant amortized time cost.
ensure capacity does not increase the size
 
Thomas Owens
Thomas Owens
That's dumb as hell.
Actually, I suppose what's worse is that the add(int, E) checks size and not capacity.
That's what should happen. But by that point, you might as well make add(int, E) just add the appropriate elements to fit.
 
ratchet freak
ratchet freak
you can do list.addAll(Collections.nCopies(desiredSize-list.size(),null));
 
Thomas Owens
Thomas Owens
Yeah. It's still fairly easy to achieve.
 
durron597
durron597
3:55 PM
you definitely can't do arrayList.ensureCapacity(10);, arrayList.set(9, foo);.
 
Thomas Owens
Thomas Owens
@durron597 Yeah. You should be able to do that or just arrayList.set(9, foo) and have it resize.
There's no reason why you shouldn't be able to.
 
durron597
durron597
I think that throwing an OutOfBounds exception is the obvious behavior, and not doing that is the outlier case
 
Thomas Owens
Thomas Owens
It makes sense for accessing. How does it make sense for inserting into a list that is capable of holding null?
 
durron597
durron597
set should affect one and only one cell.
 
Thomas Owens
Thomas Owens
@durron597 Set, maybe. Add, no.
 
durron597
durron597
3:57 PM
@ThomasOwens Add should also affect one and only one cell!
If it's going to affect more than one cell, the method name should reflect it, e.g. addAll
 
Thomas Owens
Thomas Owens
This is effectively saying that set(int, E) and add(int, E) are the same thing. They shouldn't be.
 
durron597
durron597
they aren't the same thing
 
ratchet freak
ratchet freak
add will shift the subsequent cells further
 
durron597
durron597
@ratchetfreak ^^ this
 
Thomas Owens
Thomas Owens
@ratchetfreak Yep. That means if affects more than one cell.
 
durron597
durron597
3:59 PM
oh, I was talking about add(E) not add(int, E)
 
Thomas Owens
Thomas Owens
@durron597 Yeah. add(E) just adds to the end. That's OK. add(int, E) and addAll(int, Collection) should allow you to specify an index that is beyond the end of the current list.
 
durron597
durron597
I think having null in a list is rarely desired behavior.
I assume that's probably why.
 
Thomas Owens
Thomas Owens
Even more, if you give toArray() an array that is bigger than the current list, then it will add nulls to the end of that array.
 
durron597
durron597
@ThomasOwens No it doesn't
Arrays are full of null by default when they're created
1. create array 2. insert elements into array
there's no "add nulls" step
 
Thomas Owens
Thomas Owens
> If the list fits in the specified array with room to spare (i.e., the array has more elements than the list), the element in the array immediately following the end of the collection is set to null. (This is useful in determining the length of the list only if the caller knows that the list does not contain any null elements.)
 
durron597
durron597
4:04 PM
@ThomasOwens huh.
I stand corrected
 
Thomas Owens
Thomas Owens
There's quite a bit of inconsistency there. I think I'm going to write my own ArrayList<E> now because ArrayList sucks.
 
enderland
enderland
@ThomasOwens if you don't have a different capacity/size, then every time you add a new element all existing elements need to be copied into a new array object behind the scenes
 
Thomas Owens
Thomas Owens
@enderland That makes sense.
But why does add() check the size and not the capacity? You can increase the capacity based on fullness.
Wonder what I should call my new and improved ArrayList?
 
enderland
enderland
I expect there is something like: if elements.size() < arraylist.capacity()-1 then createNewArrayWithMoreCapacity
 
Thomas Owens
Thomas Owens
Probably. But don't hashtables use a fullness ratio to determine when to resize?
 
user55340
4:10 PM
Grepcode has the code.
 
Ampt
Ampt
@MichaelT Sorry, I was busy vacationing in a place you've probably never heard of.
 
user55340
 
Ampt
Ampt
Also, HAPPY COFFEE DAY!
@AshleyNunn just finished reamde... I don't even know how to feel about that ending
1000 pages of buildup for 40 pages of action that ends with a friggin mountain lion?
 
Thomas Owens
Thomas Owens
> When the number of entries in the hash table exceeds the product of the load factor and the current capacity, the hash table is rehashed (that is, internal data structures are rebuilt) so that the hash table has approximately twice the number of buckets.
 
ratchet freak
ratchet freak
@ThomasOwens because size is a property of the List interface and capacity is an implementation detail that got exposed to let programmers minimize allocations
 
Thomas Owens
Thomas Owens
4:12 PM
So I guess I'm making a DynamicArray<E> that uses the concept of a load factor.
 
user15026
@Ampt I know, right?
 
Ampt
Ampt
and so many back and forths between characters that alternately were and weren't in combat
just so that he could avoid the omniscient point of view I guess
 
Jimmy Hoffa
Jimmy Hoffa
I couldn't help it..
 
Ampt
Ampt
rolls up newspaper
Bad Jimmy!
 
Robert Harvey
Robert Harvey
Very few unit tests in this thing.
 
Jimmy Hoffa
Jimmy Hoffa
4:20 PM
Last night's Netflix session: Starry Eyes was I think trying to be some metaphorical thing a bit more than necessary, but at the end of the day it was some slasher which didn't make sense... Well put together but bizarre and random, and it ends without any explanation. It was all about the atmosphere, sense of foreboding, and eventual inexplicable madness and murder... Blah. Way too damned slow also.
 
Robert Harvey
Robert Harvey
Peyote does that to you.
 
Jimmy Hoffa
Jimmy Hoffa
@RobertHarvey I will forever associate it with Owen Wilson in The Royal Tennenbaums
 
Thomas Owens
Thomas Owens
I'm wondering if subclassing ArrayList is the right thing to do. Maybe I should write my own implementation of the List interface?
 
Robert Harvey
Robert Harvey
Gotta not see that movie.
@ThomasOwens Why not.
Note that the List implementations typically have an Array under the hood.
 
Thomas Owens
Thomas Owens
@RobertHarvey I guess I'll only be changing size(), addAll(int, Collection<E>), add(int, E).
 
Jimmy Hoffa
Jimmy Hoffa
4:23 PM
 
Thomas Owens
Thomas Owens
That's not a whole lot of changes, and I get a ton of reuse.
 
durron597
durron597
@ThomasOwens @ratchetfreak Wow... stackoverflow.com/questions/27644361/…
 
Thomas Owens
Thomas Owens
Maybe isEmpty? I haven't decided if isEmpty should return true if all elements are null.
@durron597 CheckedExceptions suck. Can they go away yet?
 
durron597
durron597
@ThomasOwens I was on their side until I read that question.
That like... shits over the entire concept of java 8 streams.
 
Thomas Owens
Thomas Owens
@durron597 You actually liked checked exceptions?
 
Robert Harvey
Robert Harvey
4:27 PM
All that trouble, and he probably shouldn't be bothering with checked exceptions at all.
 
durron597
durron597
@ThomasOwens IDEs make them easy
 
Thomas Owens
Thomas Owens
Whenever I write a custom exception, I always make it a subclass of RuntimeException.
 
durron597
durron597
I would probably hate them if I had to write code in notepad.
 
Thomas Owens
Thomas Owens
@durron597 IDEs (Eclipse, anyway) makes unchecked exceptions easy, too.
 
Ampt
Ampt
Always a subclass of runtime?
really?
 
Thomas Owens
Thomas Owens
4:29 PM
Or a subclass of RuntimeException.
RuntimeException is the parent class of all unchecked exceptions.
 
Ampt
Ampt
so you never give explicit exception declarations?
 
Thomas Owens
Thomas Owens
Or do I have the wrong class?
 
Ampt
Ampt
as in "this method can throw an exception"?
 
Thomas Owens
Thomas Owens
@Ampt It's in the doc.
 
Ampt
Ampt
I know
but you're now hiding your exceptions and where they get thrown
instead of using the throws declaration
 
user55340
4:30 PM
@Ampt you might be surprised at my familiarity with vacation locations. And I still need to figure out how to make a bug tracker name "PBR"
 
Ampt
Ampt
Public Bug Repository?
 
Robert Harvey
Robert Harvey
Hasn't the checked exceptions debate been had already? Checked exceptions lost.
 
Ampt
Ampt
Perfect Bug Repository?
 
durron597
durron597
@ThomasOwens I basically agree completely with Josh Bloch in Effective Java item 58
 
Thomas Owens
Thomas Owens
@Ampt If I could indicate throws and list runtime exceptions, I would.
But I can't. So I rely on the javadoc.
 
Ampt
Ampt
4:32 PM
@RobertHarvey I didn't realize that we were deciding the future of programming in here. Forgive me for not conceding my ways instantly, oh wise and powerful one.
 
Thomas Owens
Thomas Owens
I let the caller or user of the API know if they can handle it or not.
 
Jimmy Hoffa
Jimmy Hoffa
@Ampt isn't that what maven is?
 
Ampt
Ampt
Yeah, but maven doesn't shorten down to PBR, duh.
 
Thomas Owens
Thomas Owens
Gah. In ArrayList, I have no access to anything useful. Like elementData.
 
Jimmy Hoffa
Jimmy Hoffa
@MichaelT Problem Bulletin Reference
 
MetaFight
MetaFight
4:42 PM
coffee day
 
Robert Harvey
Robert Harvey
@Ampt [shrug] Anders Hejlsberg looked at including checked exceptions in C#, but decided they were more trouble than they were worth.
 
Jimmy Hoffa
Jimmy Hoffa
@Ampt the thing is, checked exceptions are a cool idea, that years of practice have shown everybody in Java just works around rather than utilizing. I suppose the real question is: Should the work arounds be removed, or the checked exceptions? I think for the type of code and coder it's gear for, the checked exceptions should be removed.
That's pretty much what everyone seems to agree on for the most part. If the language had any focus on correctness, then I would be all for checked exceptions. But the language has no sense of damage control or correctness validation at the language level, so they just end up getting in the way of people who chose the language to avoid worries about correctness.
 
Thomas Owens
Thomas Owens
@JimmyHoffa Do you think that there is a language that should have checked exceptions based on the type of code and coder it's geared for? Or maybe what code and coder would drive the need for checked exceptions?
 
Jimmy Hoffa
Jimmy Hoffa
@ThomasOwens it's not a need, it's a feature. It's a simple matter of declaring all possible output types from a method - including the exceptional output types, as a part of the method type signature. Various functional programming languages already have that. It makes sense when you want to have stricter guarantees in your code, perhaps you're doing really critical stuff like aerospace/healthcare software- guarantees are good then.
 
Thomas Owens
Thomas Owens
@JimmyHoffa When it comes to guaranteed correctness, wouldn't design by contract be more useful than checked exceptions?
 
Jimmy Hoffa
Jimmy Hoffa
4:55 PM
@ThomasOwens it's the same difference - methods can error, so forcing them at compile time to declare as a part of their signature their error results before allowing them to generate such is design by contract.
For all the endless reams of Java coders writing internal LOB apps for their marketing dept etc, correctness matters very little, so getting them moving faster and allowing errors is really fine.
 
Robert Harvey
Robert Harvey
So it's all about code as documentation then.
 
Thomas Owens
Thomas Owens
I would like a way to be able to specify that my method throws a runtime exception.
 
Jimmy Hoffa
Jimmy Hoffa
@RobertHarvey no, it's about compilers guaranteeing contracts
@ThomasOwens and?
 
Thomas Owens
Thomas Owens
That would make it easier to be documentation and not relying on Javadoc. It would also make it easier for IDEs or static analysis tools to throw up warnings or errors if you fail to handle an exception.
 
Jimmy Hoffa
Jimmy Hoffa
The sad fact is Java is used a lot in aerospace, healthcare, and other spaces where criticality matters
 
Robert Harvey
Robert Harvey
4:57 PM
Healthcare, maybe. Aerospace, not so much.
Unless it's aerospace HR.
 
Thomas Owens
Thomas Owens
Yeah, we don't use Java for critical apps. Test tools and user facing stuff, mostly. The actual critical stuff tends to be C, C++, or even Ada.
Which may or may not be scarier.
 
Jimmy Hoffa
Jimmy Hoffa
@ThomasOwens yeah.. though the reasons are understandable. That said, I'm sure Java is used in Aerospace in some places most likely. Though yes, more healthcare than aerospace...
@ThomasOwens So you're arguing checked exceptions are a good thing, right?
 
Thomas Owens
Thomas Owens
@JimmyHoffa No.
 
Jimmy Hoffa
Jimmy Hoffa
I guess I don't understand what you're arguing...
 
Thomas Owens
Thomas Owens
No language should force me, by default, to handle any exception. However, I should be able to use my code as documentation and allow declared exceptions to throw a warning if they aren't handled.
 
whatsisname
whatsisname
5:00 PM
if you think Java being used in aerospace is frightening or sad, Java is a great step up
compared to the amount of stuff programmed with Simulink
 
Thomas Owens
Thomas Owens
So I can have public int foo() throws RuntimeException as a valid statement. In a default configuration, if I didn't catch RuntimeException in a caller, that's no problem. However, I could write tools or toggle IDE or compiler settings to turn not catching the exception into a warning or error.
 
Jimmy Hoffa
Jimmy Hoffa
@ThomasOwens If it's critical applications, I would say - the language definitely should force you to handle all possible errors. How you handle them is up to you, not all can be corrected but you should definitely halt the error from causing catastrophic failure. I would like the compiler to force coders to do that in the event their software may have impacts on someone's health
 
Thomas Owens
Thomas Owens
@JimmyHoffa Ideally, that's nice. But if you were building a general purpose language (Java), why would you force that on everyone?
If you were building a special language for critical software, yeah, you're right.
 
Jimmy Hoffa
Jimmy Hoffa
@ThomasOwens you wouldn't! My point exactly is - checked exceptions don't belong in Java because it's not geared towards the scenarios I refer to above.
 
Thomas Owens
Thomas Owens
@JimmyHoffa I'd say that checked exceptions don't belong in most languages. But the ability to support tools that facilitate more robust software should be part of the language.
It turns the language into an ecosystem of not only the compiler, but IDEs and static and dynamic analysis tools.
 
whatsisname
whatsisname
5:04 PM
There are missile systems in active use by the US that have their control systems programmed in simulink.
 
Thomas Owens
Thomas Owens
@whatsisname Simulink is used in a lot of airplanes and automobiles.
I've never used it. What's so bad about it?
 
whatsisname
whatsisname
it's a lot like labview
but with more matlab
 
Thomas Owens
Thomas Owens
From what I can tell, the model based development allows for a lot of simulation without hardware to ensure that your Simulink models are accurate and correct before you put it on hardware for integration, SIL, flight test, and production.
 
whatsisname
whatsisname
but its a box->line type programming thing
it's one of those things that's really neato for simple models and systems
but gets to be really nasty when you do a real-world complexity project
 
Jimmy Hoffa
Jimmy Hoffa
@whatsisname and ones written in Java. Look, I know bureaucracy is hardly the solution to choosing or designing anything with technical quality, and whenever you're talking about critical life affecting things bureaucracy is bound to be the first tool that get's used. So realistically, all of these types of systems are plausibly some of the worst... scary thought, but likely true
 
whatsisname
whatsisname
5:07 PM
and the box->line system is terrible at resolving timing issues and other aspects of the physical hardware
 
Thomas Owens
Thomas Owens
@whatsisname If you decompose the problems into small enough pieces, is anything really that complex?
 
whatsisname
whatsisname
yes
 
Thomas Owens
Thomas Owens
@whatsisname That is a true statement - I can't argue that. But that's what lab integration and SIL are for.
 
Jimmy Hoffa
Jimmy Hoffa
@ThomasOwens it's all just 1s and 0s after all
 
whatsisname
whatsisname
especially devices where you are on the knifes edge of cost and performance
the real world is an inherently complex place and a device must ultimately address every aspect of that complexity
no abstraction can completely cover it
 
Thomas Owens
Thomas Owens
5:09 PM
That's a true statement.
But then again, your operating tolerances can reduce the complexity, can't they?
 
whatsisname
whatsisname
and being on that knife edge is where stuff like missiles hangs out
 
Thomas Owens
Thomas Owens
Yeah.
 
whatsisname
whatsisname
programming an air conditioner with simulink? yeah sure go ahead noones going to care
 
durron597
durron597
@JimmyHoffa "Everybody in Java" not everyone.
 
whatsisname
whatsisname
so simulink suffers from two problems: 1 is it doesn't provide the details to go straight to the hardware that is occasionally needed
and 2: it is another one of those "programming for non-programmer" platforms
those two things combined = terror
 
Thomas Owens
Thomas Owens
5:13 PM
I do agree with 2. I don't like moving software development out of the hands of people with years of education and experience developing software. There are a lot of good control systems engineers out there, but I'd rather have software engineers build software from their models and algorithms.
 
Jimmy Hoffa
Jimmy Hoffa
@durron597 enough of them that overall it's providing more wasted time and mental space than utility
 
whatsisname
whatsisname
-2
Q: Can my client program always see a local IP address?

Jim BeamI have written a client program that is limited to certain IP addresses. I check the IP addresses on the server side. Of course, someone can use a proxy server and access my server with a "valid" IP address. So I am thinking about adding some client-side code to check the local IP address and ma...

security code-security
OP is insulting commenter, temptation to insult back is high
 
durron597
durron597
 
enderland
enderland
5:33 PM
@whatsisname heh his comment about "you r dumb" was deleted by my single flag, that means 5 other people had flagged it lol
 
Thomas Owens
Thomas Owens
It would be nice if Java included a basic suite of test cases.
 
enderland
enderland
so in other news, I think I'm going to scrap my entire thesis and fight to drop my adviser, and do a capstone project instead... I should write a book about this grad school experience
 
Thomas Owens
Thomas Owens
For example, there's an AbstractList<E> abstract class that includes inner classes. Rather than just me inspecting their code to see if it's useful, let me run test cases against my implementation of the interface and see red/green.
 
gnat
gnat
@ThomasOwens how about JCK? it cost money though
 
durron597
durron597
@ThomasOwens I flagged the question that @whatsisname linked above
 
Thomas Owens
Thomas Owens
5:37 PM
Yeah...I'm looking at it.
 
enderland
enderland
probably quite a few flags there, that chap is quite the tool
 
Thomas Owens
Thomas Owens
@gnat Isn't that for someone implementing their own Java compiler or JDK?
 
durron597
durron597
@enderland I "custom"ed saying basically "nuke all plz"
 
enderland
enderland
Mine was something similar
 
gnat
gnat
@ThomasOwens yes. I just figured that you're talking about applications, that rules it out
 
Thomas Owens
Thomas Owens
5:38 PM
I'm thinking more for people who extend or implement the interfaces and abstract classes that come with the language to ensure that you know if you're violating the intended contract.
I wouldn't weigh them as much as my unit tests, but it would be nice to know where I am or am not following the intention of the interface
For example, the Iterator in AbstractList makes certain assumptions. I'm reading that code myself now to see if it holds true for my code. I'd rather run a test/
 
gnat
gnat
@ThomasOwens ...on a further thought, I bet one could find such cases in JCK. It would be an overkill to pay for seeing or using their code though just for that. There is a (slim) chance that Oracle tries to monetise by "spinning off" part of such tests in some other commercial product, I am too lazy to check their commercial offerings in this area
 
Thomas Owens
Thomas Owens
Yeah. I'm not going to both to look for this, since there's no way I can buy it,
It would be nice to include these things as part of the JDK, though. The ability to build your own JDK should be monetized, and I have no issues with that.
I wonder if it would be useful if I went through all of the interfaces and abstract classes in the JDK and write JUnit tests based on the descriptions of the interfaces (and maybe vetted them against OpenJDK).
 
gnat
gnat
@ThomasOwens I bet these are included. IBM builds and sells their own JDK, and they pay Oracle for certifications. And someone else does, I don't remember. Apple did at some time back, but they couldn't make it from java 5 to 6 so they dropped
 
Thomas Owens
Thomas Owens
@gnat Included in the JDK? Or the JCK?
 
gnat
gnat
@ThomasOwens IBM (and some others, and Apple in the past) build their JDK, or JRE, from scratch or something like that. Heck when Java was Sun's, Oracle themselves did their own JRE or JDK, JRockit. After they think they're ready for beta, they pay to Oracle to run the JCK (I don't know exact details, whether Oracle runs it or they themselves). They fix failures reported by failed JCK tests, and after all tests get green, they proudly stick that Java label on their stuff
 
Thomas Owens
Thomas Owens
5:51 PM
Yeah. That's overkill for testing your implementation of AbstractList<E>, though.
 
gnat
gnat
@ThomasOwens absolutely. I bet tests for that in JCK are near perfect - how couldn't they, after so many years of testing so many implementations - but buying tests for all the Java just to get several ones for AbstractList would be 200% overkill
2000%
 
Thomas Owens
Thomas Owens
But there could be value in writing test cases for a set of the interfaces and abstract classes included in Java that people may extend. The idea being that if you implement one of these interfaces or extend one of those abstract classes, you want to make sure that if you give your implementation to a class that accepts the interface or abstract class, it will conform to the expected behavior as defined by the JavaDocs.
But I don't even know how feasible that is. A lot of the methods would rely on access to inner data to test them.
 
04:00 - 18:0018:00 - 00:00

« first day (1802 days earlier)      last day (3192 days later) »