|
'welp this api is dumb and doesn't conform to its own spec but we're not gonna fix it because the primary consumer is us.'
|
#
?
Dec 11, 2015 17:03
|
|
|
|
| # ? Jun 11, 2024 13:06 |
|
|
three weeks later the private api becomes public a few years after that and it's underpinning a gigantic software ecosystem
|
|
#
?
Dec 11, 2015 17:16
|
|
|
I was looking at integration for athenahealth (another emr) and while they have a pretty good api + open documentation they also have this: http://www.athenahealth.com/cmp/more-disruption/more-disruption.php lol
|
|
#
?
Dec 11, 2015 17:20
|
|
|
Shaggar posted:I was looking at integration for athenahealth (another emr) and while they have a pretty good api + open documentation they also have this: http://www.athenahealth.com/cmp/more-disruption/more-disruption.php lol Their CEO is a big ole silicon valley dick head too. There's plenty of video of him at various conferences implying his competitors' leadership are dumb, or idiots, and at one point he implied everyone else was a "dinosaur", which is just startup code for "well established in this market and solving problems we haven't 'disrupted' yet."
|
|
#
?
Dec 11, 2015 17:23
|
|
|
they should accelerate Turing pharmaceutical out of the market but probably they'll just partner
|
|
#
?
Dec 11, 2015 17:25
|
|
|
they list already integrated partners on their site which is pretty cool though. it would be useful when submitting rfps for Athena customers.
|
|
#
?
Dec 11, 2015 17:34
|
|
|
is this the thread where i post this https://github.com/dotnet/coreclr/issues/2285
|
|
#
?
Dec 11, 2015 17:35
|
|
|
the talent deficit posted:i use and love elm. your work is much appreciated what's your elm workflow btw, i'm curious
|
|
#
?
Dec 11, 2015 17:40
|
|
|
we have an internal protocol that uses HTTP chunked encoding to delimit JSON objects. It is a piece of garbage and I am constantly angry at it.
|
|
#
?
Dec 11, 2015 17:43
|
|
|
As a Millennial I posted:is this the thread where i post this https://github.com/dotnet/coreclr/issues/2285 This is why I stopped working on open-source.
|
|
#
?
Dec 11, 2015 17:47
|
|
|
As a Millennial I posted:is this the thread where i post this https://github.com/dotnet/coreclr/issues/2285 love the way the guy keeps dangling "well, MAYBE I WONT HELP U" with the expectation that the thread will immediately rend their garments that this amazing talent is slipping through their fingers
|
|
#
?
Dec 11, 2015 18:07
|
|
|
for me it's the way he completely rejects the idea that he could be wrong, because his is clearly the objectively correct opinion and is beyond discussion
|
|
#
?
Dec 11, 2015 18:11
|
|
|
Symbolic Butt posted:didn't you contribute to the ruby parser and talked to matz and stuff? or were you just studying it? i was asking questions on behalf of someone else asking about it i did learn what EXPR_MID was though
|
|
#
?
Dec 11, 2015 18:14
|
|
|
JawnV6 posted:love the way the guy keeps dangling "well, MAYBE I WONT HELP U" with the expectation that the thread will immediately rend their garments that this amazing talent is slipping through their fingers quote:All what you are saying doesn't make any sense! You are talking about applications. You are not facing the problem as a programmer, but as a user! i originally thought he was arguing that breaking the api is worth it, but he apparently doesn't understand what "breaking the api" even means quote:You seriously think that if Decimal.Parse fails in certain situation, all the applications relying on it would fail too?! Only the applications built by incompetent programmers would fail. Competent programmers create applications which will always work independently upon virtually anything else (e.g., the errors of the given programming language, how clueless the given user might be, etc.)!
|
|
#
?
Dec 11, 2015 18:23
|
|
|
fart simpson posted:what's your elm workflow btw, i'm curious i'm using vscode with the elm plugin, elm-reactor and manually refreshing firefox i'm still relatively new to elm but i built an internal interface to our database management tools with elm + python/flask and it was basically the greatest experience of my professional career
|
|
#
?
Dec 11, 2015 18:43
|
|
|
As a Millennial I posted:is this the thread where i post this https://github.com/dotnet/coreclr/issues/2285 oh well that sounds like a reasonable thing to discu-- quote:Ey, stupid: I can laugh at you (at your ignorance and at your lack of capability to understand even the simplest idea) and let you think that you have everything under control. But, piece of poo poo, don't ever dare to insult me on my face. 
|
|
#
?
Dec 11, 2015 19:17
|
|
|
Shaggar posted:http://www.asp.net/ How;s ASP.NET doing on non-windows platforms? I've pretty much moved off of Windows at this point but drat if I don't miss C# especially LINQ.
|
|
#
?
Dec 11, 2015 19:36
|
|
|
Ochowie posted:How;s ASP.NET doing on non-windows platforms? I've pretty much moved off of Windows at this point but drat if I don't miss C# especially LINQ. mono has a reasonably complete implementation but i've never heard of anyone using it for real work. the official microsoft .NET and ASP.NET stacks have been ported to linux but they are very much alpha quality right now. a cool tech demo not a finished product.
|
|
#
?
Dec 11, 2015 19:52
|
|
|
Notorious b.s.d. posted:mono has a reasonably complete implementation but i've never heard of anyone using it for real work. that sounds about 50% more complete than most rails-inspired tech stacks
|
|
#
?
Dec 11, 2015 20:05
|
|
|
Dylan16807 posted:oh well that sounds like a reasonable thing to discu-- this reads like a dril tweet
|
|
#
?
Dec 11, 2015 20:11
|
|
|
As a Millennial I posted:is this the thread where i post this https://github.com/dotnet/coreclr/issues/2285 He's right though. 1,2 equalling 12 is dumb.
|
|
#
?
Dec 11, 2015 21:07
|
|
|
imagine working with this cool guyquote:@MartinJohns I was (kind of) sarcastic too. With "kind of" I meant that I got your idea (= how this should work), but the reality (= what I have seen here, mainly today) doesn't seem to prove it (= lots of people wasting time, getting offended and trying to offend; just a few trying to understand the underlying problem and trying to help). In any case, thanks for your contribution (I love Dr. Who).
|
|
#
?
Dec 11, 2015 21:10
|
|
|
pointsofdata posted:He's right though. 1,2 equalling 12 is dumb. lol
|
|
#
?
Dec 11, 2015 21:14
|
|
|
He starts from such a sane place ("1,2 parsing to 12 is weird"), then goes off the rails immediately. He could propose adding a function, say something like tryParseStrict, or suggesting adding an overload with a DelimiterCulture argument, or any other number of solutions that doesn't break other people's code, but he's dead set that anyone who did depend on "1,2"=12 is an idiot so gently caress their code.
|
|
#
?
Dec 11, 2015 21:20
|
|
|
LeftistMuslimObama posted:He starts from such a sane place ("1,2 parsing to 12 is weird"), then goes off the rails immediately. He could propose adding a function, say something like tryParseStrict, or suggesting adding an overload with a DelimiterCulture argument, or any other number of solutions that doesn't break other people's code, but he's dead set that anyone who did depend on "1,2"=12 is an idiot so gently caress their code. It took my several posts before I went woah he's mental
|
|
#
?
Dec 11, 2015 21:34
|
|
|
LeftistMuslimObama posted:He starts from such a sane place ("1,2 parsing to 12 is weird"), then goes off the rails immediately. He could propose adding a function, say something like tryParseStrict, or suggesting adding an overload with a DelimiterCulture argument, or any other number of solutions that doesn't break other people's code, but he's dead set that anyone who did depend on "1,2"=12 is an idiot so gently caress their code. I think he has a better argument with the fact that "1,,,,,," parses to 1 but honestly I fail to see how ignoring the grouping delimiter isn't always the sane thing to do. One question I had reading that thread, how does .NET currently treat locale-specific grouping delimiters (e.g. In many countries '.' is the group delimiter). Edit: The guy is batshit crazy though. That was a pretty funny read. Also, I feel like this guy read some of Linus's rants and felt they're something he should emulate. Edit #2: Also he completely fails to grasp the difference between a group delimiter and the decimal delimiter. Ochowie fucked around with this message at 21:43 on Dec 11, 2015 |
|
#
?
Dec 11, 2015 21:36
|
|
|
Ochowie posted:I think he has a better argument with the fact that "1,,,,,," parses to 1 but honestly I fail to see how ignoring the grouping delimiter isn't always the sane thing to do. One question I had reading that thread, how does .NET currently treat locale-specific grouping delimiters (e.g. In many countries '.' is the group delimiter). You can pass it a culture which deals with that sort of thing
|
|
#
?
Dec 11, 2015 21:43
|
|
|
Ochowie posted:I think he has a better argument with the fact that "1,,,,,," parses to 1 but honestly I fail to see how ignoring the grouping delimiter isn't always the sane thing to do. One question I had reading that thread, how does .NET currently treat locale-specific grouping delimiters (e.g. In many countries '.' is the group delimiter). True. Either way, if you want "1,,,,," to return a failure instead of 1, the right thing to do is either write a new function that enforces strict grouping or add an additional locale-sensitive argument to the existing function in such a way that it doesn't break existing code.
|
|
#
?
Dec 11, 2015 21:44
|
|
|
GrumpyDoctor posted:You seriously think that if Decimal.Parse fails in certain situation, all the applications relying on it would fail too?! Only the applications built by incompetent programmers would fail. Competent programmers create applications which will always work independently upon virtually anything else (e.g., the errors of the given programming language, how clueless the given user might be, etc.)! oh yeah this is definitely the best part
|
|
#
?
Dec 11, 2015 21:46
|
|
|
LeftistMuslimObama posted:True. Either way, if you want "1,,,,," to return a failure instead of 1, the right thing to do is either write a new function that enforces strict grouping or add an additional locale-sensitive argument to the existing function in such a way that it doesn't break existing code. Oh, I totally agree. I actually think parsing "1,,,," to "1" is the most sane thing to do it just looks really weird. pointsofdata posted:You can pass it a culture which deals with that sort of thing I figured it was something like that but it's been a while.
|
|
#
?
Dec 11, 2015 21:50
|
|
|
Ochowie posted:Oh, I totally agree. I actually think parsing "1,,,," to "1" is the most sane thing to do it just looks really weird. Yeah. It's the right thing to do until it's not, which is why there should be a separate function for strict behavior. If I had his use case, I'd probably just write my own wrapper around tryParse that does the grouping checking (probably using a regex?) and then call tryParse itself.
|
|
#
?
Dec 11, 2015 22:06
|
|
|
I agree with the guy that 1,,,,,,5 should probably not be considered a valid representation of 15, but as people have said the change should probably be implemented by introducing a new ParseStrict function or whatever.
|
|
#
?
Dec 11, 2015 22:11
|
|
|
What I'm wondering, is it better to have a parseStrict() included in the language or is it smarter to tell users to implement it themselves? Having a parseStrict() to paper over an older, probably not correct, function gives me strong mysql_real_escape_string vibes. You could put a notice in the docs of parseStrinct() saying that it probably doesn't do what you want it to, and everyone needs to reimplement the more correct function themselves. I'd probably would go for the latter option in this case.
|
|
#
?
Dec 11, 2015 22:20
|
|
|
Not to mention it's a parser, not a validator. If you need it to fit a specific format, validate the input. Plenty of integer parsers will parse "154foobar" as 154 and i've never heard anyone complain of that. Hell I'm happy if it parses "foo" to 0. If I actually care that the input matches a sane definition of valid (as opposed to just a sufficient one), then I'll do extra validation. I've had many arguments with people like this. Usually it's first- or second-year students who have attained enough confidence to know the black-and-white rules about what you should and shouldn't ever (EVER) do in your code, without any of the experience or rationale to back any of it up. Someone with roughly one fifteenth of my programming experience (just in terms of years) got very passionate and livid that I was okay with using multiple return statements in a single function, and told me that I was being irresponsible. See also: Rob Pike's weird opinions about pretty much everything. edit: Considering it's a lot easier to constrain the behaviour of Parse by wrapping it than it is to relax its strictness after-the-fact without completely reimplementing it, it's not even a question that should be asked. Next we're going to be complaining about HTML parsers that accept malformed HTML. Volte fucked around with this message at 22:33 on Dec 11, 2015 |
|
#
?
Dec 11, 2015 22:28
|
|
|
i didn't come here to have this discussion. i've already pointed out that it's wrong and i can implement it myself so please just tell me how to do that and i will fix it
|
|
#
?
Dec 11, 2015 22:35
|
|
|
KARMA! posted:What I'm wondering, is it better to have a parseStrict() included in the language or is it smarter to tell users to implement it themselves? Having a parseStrict() to paper over an older, probably not correct, function gives me strong mysql_real_escape_string vibes. You could put a notice in the docs of parseStrinct() saying that it probably doesn't do what you want it to, and everyone needs to reimplement the more correct function themselves. I'd probably would go for the latter option in this case. I'd say that the right answer is having users write their own wrapper. I can't think of many situations off-hand where you'd be putting input like that into tryParse (maybe if it was end-user input, but I imagine your front-end is doing some validation on the user input first unless it's like a console utility), so if it genuinely comes up for you you should write it yourself to ensure it truly fits your specific use-case. Every consumer is likely to have a different idea of what strictParse should be strict about, so trying to accommodate everyone's edge cases in library code is a path to a big mess. Maybe I want 1,,,,,, to be valid input, but 1,,,,,,,,. not to be, but maybe I want it the other way around.
|
|
#
?
Dec 11, 2015 22:38
|
|
|
1,,,,,,5 being 15 seems like some goofy euler problem where you're finding bigints with some reduction down to other properties if he's so convinced it's a technical problem, i'd poke at understanding if the existing cultural hooks are rich enough to describe the Indian comma formatting
|
|
#
?
Dec 11, 2015 22:44
|
|
|
As a Millennial I posted:oh yeah this is definitely the best part haha, competent programmers make things work independent of virtually anything. look Ma, I'm a loving wizard, no dependency changes can harm me.
|
|
#
?
Dec 11, 2015 22:56
|
|
|
guy is certainly right that it was a sort of bad implementation choice, but i hold it pretty likely that it exists for some good reason (windows having a locale thing for those separators but ancient myanmar script having a rule that they should be placed at every prime number of digits or some poo poo, a locale piece of information the system did not export and which was horrific to recreate, so, whatever, just permit them everywhere). it certainly can't be changed loving peoples code over (if the value of the code implementing the platform is not a tiny fraction of the value of the code running on the platform you are a failure and should just give up anyway), and, more than anything, no one, except perhaps yospos, should care enough to write more than 10 words on the matter. if you're going to type hundreds of words you can write your own parser with less effort
|
|
#
?
Dec 11, 2015 23:11
|
|
|
|
| # ? Jun 11, 2024 13:06 |
|
|
lol if your code would break simply because a core platform api suddenly started rejecting input that was previously stated to be valid in the formal documentation, just lol
|
|
#
?
Dec 11, 2015 23:19
|
|