|
Asymmetrikon posted:nobody praises java's syntax, it's a bog standard ALGOL-like yeah. it's boring, stable, and has a library for everything. the language itself is even getting pretty good since java 8.
|
# ? Mar 1, 2017 18:27 |
|
|
# ? May 22, 2024 07:46 |
|
Nobody praises Java, but it gets the job done reliably. It's unexciting and doesn't foist many surprises on you. The tedium of working with it is also massively overstated by its detractors. Its community also has a fairly strong consensus on how to get poo poo done. One Java codebase looks much like any other, especially if it uses Google's frameworks which afaik are becoming fairly ubiquitous these days. So you don't have to indoctrinate a new contributor into your organization's house variant of whatever language you're using, they can jump in, start exploring the relationships between types, and get straight to work.
|
# ? Mar 1, 2017 18:59 |
Doom Mathematic posted:A code snippet which properly supported human names would be appallingly complex no matter what programming language it was written in. Yeah but you could do a lot better by combining the "first name" and "surname" fields into just one "name" field in the JSON. Edit: picking on the name thing was stupid though, and I should have just left it at "that Haskell is actually pretty readable".
|
|
# ? Mar 1, 2017 19:02 |
|
Sapozhnik posted:One Java codebase looks much like any other, especially if it uses Google's frameworks which afaik are becoming fairly ubiquitous these days. one of the problems i think is that people my age learned java almost a decade ago. back then it was not very good, not just from a language standard but from a community standard. times have changed though, for the better.
|
# ? Mar 1, 2017 19:30 |
|
I decided long ago that I just have a 'name' field and an 'address' field. They're both too complicated for my weak mind.
|
# ? Mar 1, 2017 19:31 |
|
java will never cause me grief because of anything in the nature of the thing itelf. the implementation and ecosystem are well-known and rock solid, you can scale and monitor and generally *work* with it, it neither falls in the trap of being a configurable mess where every site has a special snowflake setup, nor is it a black box that cannot be twisted a bit if the need arises more fundamentally than that though; server java development is the one role in which i think even i am competent enough to hire some programmers for with a reasonable expectation that the right thing will eventually get built, and that what is built is likely to be something that can to some extent be maintained and tweaked (this to a great extent is down to the standard style of java enforcing very small units of abstractions which makes it possible for less skilled programmers to get useful things done without putting the overall project at risk)
|
# ? Mar 1, 2017 19:48 |
|
Doom Mathematic posted:A code snippet which properly supported human names would be appallingly complex no matter what programming language it was written in. can confirm, our name formatting code is a behemoth, especially since we have to support dutch and danish names too.
|
# ? Mar 1, 2017 19:54 |
|
Doom Mathematic posted:A code snippet which properly supported human names would be appallingly complex no matter what programming language it was written in. See also: http://www.kalzumeus.com/2010/06/17/falsehoods-programmers-believe-about-names/
|
# ? Mar 1, 2017 20:22 |
|
Cybernetic Vermin posted:java will never cause me grief because of anything in the nature of the thing itelf. the implementation and ecosystem are well-known and rock solid, you can scale and monitor and generally *work* with it, it neither falls in the trap of being a configurable mess where every site has a special snowflake setup, nor is it a black box that cannot be twisted a bit if the need arises the jvm is one of the better aspects of the platform imo, in that the same code does run on quite a few platforms without much tweaking. i write code that i expect to run on windows, linux, several variants of unix across intel and power, as/400, and z/os, but am able to catch the vast majority of bugs just by testing regularly on linux with less frequent checks on those other platforms. java will never be great, but it's far from the worst out there
|
# ? Mar 1, 2017 20:28 |
|
Asymmetrikon posted:one of the problems i think is that people my age learned java almost a decade ago. back then it was not very good, not just from a language standard but from a community standard. times have changed though, for the better.
|
# ? Mar 1, 2017 21:26 |
|
Asymmetrikon posted:one of the problems i think is that people my age learned java almost a decade ago. back then it was not very good, not just from a language standard but from a community standard. times have changed though, for the better. thing is though, when it gets to this, that rob pike was right: you need to design a lot of programming languages with a mind towards people who will work 9-5, being professionals, but not being interested enough to be fantastically knowledgeable or up to date about every matter of it. java 1.4 was certainly not a pleasant place to be, but it was very much already in that building block zone, where many people could rather safely collaborate on some ambitious project, while individually not being too bothered when they or a colleague had to duck out early to pick up the kids overall this thread is very weird in that it tries to merge these things into one whole, because a lot of what is discussed here will never get into that situation, but a lot of code will have to get written in precisely that way
|
# ? Mar 1, 2017 21:42 |
|
ulmont posted:See also: that article is pretty drat stupid, if you ask me. outside of the obvious stuff like "use Unicode" and "don't try to do dumb validation on names", that stuff might be nice to talk about, but at some point you need to actually create a thing and I'm pretty sure that most people would rather use a program or, lets be honest, website that exists but doesn't perfectly handle their "unusual" naming conventions (just like 99.9% of things in the world) than creating the perfect website through pure force of will. Like, if you ask people to enter a separate first and last name and store them as arbitrary Unicode code points, you've done your due diligence.
|
# ? Mar 1, 2017 21:51 |
|
nah the takeaway is pretty straightforward, just have a name field that is a kilobyte blob of unicode and don't use it internally for anything important. dont try to sanitize it, dont use it as a key, dont split it up, just leave it alone
|
# ? Mar 1, 2017 21:53 |
|
i think the point of the article is that it's futile to attempt to handle names "intelligently" in most cases, and that you're better off just asking the user for a blob to call them.
|
# ? Mar 1, 2017 21:56 |
|
Sinestro posted:Like, if you ask people to enter a separate first and last name and store them as arbitrary Unicode code points, you've done your due diligence. BobHoward posted:same it's like the Underneath of x86, you'd sorta recognize things like RAX but they've been perverted and abstracted a few times
|
# ? Mar 1, 2017 22:28 |
|
JawnV6 posted:what meaningful things are you going to do with those 2 separate blobs? where does my middle name/initial go? where does "the sixth"/"VI" go? you'd have like a { first_name: "poo poo", last_name: "rear end-Fuckson VI" } kind of thing, where you have "what you use when you say "hey, <person>!" and "the other part".
|
# ? Mar 1, 2017 22:32 |
|
The problem with that article is that it's not a spec. It's just a list of terrible things which you mustn't do because it means you're a failure. It gives very few clues as to what you should actually do, and it isn't even clear that it's possible to completely satisfy anybody. Like false assumption number 11, "People’s names are all mapped in Unicode code points." What's the correct way to solve that problem? Plus, the list itself claims to be incomplete, which means that even if you satisfy everything in the list you've probably still failed for some other reason which you were too stupid to anticipate. It's weirdly passive-aggressive. Bloody posted:nah the takeaway is pretty straightforward, just have a name field that is a kilobyte blob of unicode and don't use it internally for anything important. dont try to sanitize it, dont use it as a key, dont split it up, just leave it alone Falsehood #41: a person's name takes up less than a kilobyte of Unicode. Gotcha!
|
# ? Mar 1, 2017 22:34 |
|
Bloody posted:nah the takeaway is pretty straightforward, just have a name field that is a kilobyte blob of unicode and don't use it internally for anything important. dont try to sanitize it, dont use it as a key, dont split it up, just leave it alone seriously. many many things don't need anything more than a full name (for official communications) and a preferred name (for informal communications). job done unless there's a reason to pull apart someone's name, in which case you try to match the locale standards the best you can.
|
# ? Mar 1, 2017 22:34 |
|
Doom Mathematic posted:The problem with that article is that it's not a spec. It's just a list of terrible things which you mustn't do because it means you're a failure. It gives very few clues as to what you should actually do, and it isn't even clear that it's possible to completely satisfy anybody. Like false assumption number 11, "People’s names are all mapped in Unicode code points." What's the correct way to solve that problem? Uhh that's the point though. This is a battle you cannot win. A name is a black box. It may have spaces, it may have special characters, it may have any number of parts.
|
# ? Mar 1, 2017 22:35 |
|
Sinestro posted:you'd have like a { first_name: "poo poo", last_name: "rear end-Fuckson VI" } kind of thing, where you have "what you use when you say "hey, <person>!" and "the other part". names of people are very complicated, and it is vaguely charming that they are, since it seems like it'd be tarnishing the humanity of the situation to have quick and easy fields to explain it fully to a computer notably the above rather misses the fact that in some cultures it is very contextual whether you may use the "real" first name, or whether you are on an initials basis only
|
# ? Mar 1, 2017 22:37 |
|
a lot of hacker news posters in this thread right now
|
# ? Mar 1, 2017 22:46 |
|
Sinestro posted:you'd have like a { first_name: "poo poo", last_name: "rear end-Fuckson VI" } kind of thing, where you have "what you use when you say "hey, <person>!" and "the other part". so 0/3 on the specific questions while adding some wholly new fuckups just because you want two fields instead of one opaque blob, gj Doom Mathematic posted:The problem with that article is that it's not a spec. It's just a list of terrible things which you mustn't do because it means you're a failure. It gives very few clues as to what you should actually do, and it isn't even clear that it's possible to completely satisfy anybody. Like false assumption number 11, "Peoples names are all mapped in Unicode code points." What's the correct way to solve that problem? you'll gently caress up, don't try Fergus Mac Roich posted:Uhh that's the point though. This is a battle you cannot win. A name is a black box. It may have spaces, it may have special characters, it may have any number of parts.
|
# ? Mar 1, 2017 22:48 |
|
JawnV6 posted:ucode is like grabbing the opcode wires of a 5-stage pipelined RISC and treating them as a meaningful language i'd imagine it would look like some weird stepping-specific VLIW monstrosity if anything
|
# ? Mar 1, 2017 22:52 |
|
Sinestro posted:you'd have like a { first_name: "poo poo", last_name: "rear end-Fuckson VI" } kind of thing, where you have "what you use when you say "hey, <person>!" and "the other part". we have a seperate field for dutch people to specify their van/von/whatever because like 90% of dutch people have that and they don't like it to affect alphebatization. the approach we've taken is simply to when we sign someone in a new country work with them to develop support for their name-related needs. so sure we don't have support for vietnamese names right now, but we have arabic script and name support. in an ideal world you support more name-related needs as they come, but a lot of programmers want to poo poo-it-and-forget-it with their products and never support them once they're done.
|
# ? Mar 1, 2017 22:55 |
|
The Chicago manual of style has a full section on name sorting for indexing and that stuff is batshit confusing. Here's a short extract: poo poo is hard.
|
# ? Mar 1, 2017 23:22 |
|
MononcQc posted:Programming is hard.
|
# ? Mar 1, 2017 23:45 |
|
JawnV6 posted:why do you need my name? why is an email/username insufficient? Can you seriously not think of a single instance where a human might want or need to enter their actual name into a computer?
|
# ? Mar 2, 2017 01:11 |
|
Doom Mathematic posted:Can you seriously not think of a single instance where a human might want or need to enter their actual name into a computer? avoid if possible
|
# ? Mar 2, 2017 03:52 |
|
i make the pizza delivery man refer to me by my unicode code points
|
# ? Mar 2, 2017 04:56 |
I legally changed my name to a sequence of bits which is not valid UTF-8
|
|
# ? Mar 2, 2017 06:30 |
|
let's be real here: no one in yospos is breaking into any markets where a simple first/last name pair won't suffice
|
# ? Mar 2, 2017 09:26 |
|
wouldn't surprise me to learn that most people who use the internet have figured out a way to work with the anglosphere's first/last name convention
|
# ? Mar 2, 2017 09:27 |
|
i think the correct conclusion is that if you aren't going to make an ongoing effort to handle names correctly you probably should at least not constrain or lecture people on how they should enter stuff. probably pretty reasonable to just have "full name" and "what should we call you?" without trying to interpret beyond that
|
# ? Mar 2, 2017 10:23 |
|
redleader posted:wouldn't surprise me to learn that most people who use the internet have figured out a way to work with the anglosphere's first/last name convention except for the people who only have a single name sure.
|
# ? Mar 2, 2017 11:18 |
|
I vaguely remember that some people with a single name just double it up for places that demand two
|
# ? Mar 2, 2017 11:29 |
|
then the question is why not do this?Cybernetic Vermin posted:i think the correct conclusion is that if you aren't going to make an ongoing effort to handle names correctly you probably should at least not constrain or lecture people on how they should enter stuff. probably pretty reasonable to just have "full name" and "what should we call you?" without trying to interpret beyond that because it honestly feels like white apathy/supremacy to hold on tight to first & last name as an input method unless you genuinely require further localisation of input formats, at which point you should probably still boil down to full & preferred name, with localised forms per nationality as required. if you're gonna get it structured 'right' the only way to do it is to acknowledge that one form isn't enough to do structured name input.
|
# ? Mar 2, 2017 11:54 |
|
redleader posted:I vaguely remember that some people with a single name just double it up for places that demand two ugh i hate being forced to input wrong things because whoever gently caress made the system is dumb as hell
|
# ? Mar 2, 2017 12:02 |
|
Powaqoatse posted:ugh i hate being forced to input wrong things because whoever gently caress made the system is dumb as hell i have a capital letter in the middle of my surname (McG), and for the longest time even our product couldn't handle it and would auto capitalise the M and lowercase everything else. it still can't handle the more traditional McG though e: SA can't handle the unicode superscript c either lol
|
# ? Mar 2, 2017 12:25 |
|
gonadic io posted:i have a capital letter in the middle of my surname (McG), and for the longest time even our product couldn't handle it and would auto capitalise the M and lowercase everything else. i have a dash in my name and loving google won't let me use my actual name in my email address.
|
# ? Mar 2, 2017 12:36 |
|
|
# ? May 22, 2024 07:46 |
|
I have four names. putting my legal name on american forms that expect exactly one middle initial was fun. and that's just within white anglo-saxon naming conventions.
|
# ? Mar 2, 2017 13:34 |