|
Shaggar posted:but why store it as the serialized xml? normalize that poo poo usually yes, but once in a while you can't normalize it and a serialized object column is the next best option. the alternatives are so much worse normalized data in atomic columns > serialized object column > entity-attribute-value table > half-assed homemade string concatenation and conversion mess > unspeakable horror that most closely resembles a stack-based Polish calculator for logic rules in the form of indexed sql rows NihilCredo fucked around with this message at 17:26 on Oct 11, 2017 |
# ? Oct 11, 2017 17:23 |
|
|
# ? May 15, 2024 15:53 |
|
Shaggar posted:why would you store serialized data in your database at all? sometimes you are gonna be storing full xml documents, and there are three places you can do that
storing them in the database has the advantage that now in addition to simple search/retrieval, you can also reference xml data from regular old queries that primarily access fully normalised data it's not an ideal situation but it beats the hell out of the other two options
|
# ? Oct 11, 2017 17:43 |
|
MononcQc posted:I'd be down for a new generation of web browsers that are all about static sites with no JS and very limited design capabilities. Bring back document-based web. good use of SPA - being able to manipulate and maintain page state while also loading new data. like to keep scroll position or user input. e.g. the web app for myfitnesspal would be better with more SPA features. there are forms that include paginated lists I might want to flip through for reference. but if I do I'll lose what I've entered so far in the form on the page. bad use of SPA - everything else on the web
|
# ? Oct 11, 2017 17:44 |
|
you jcan make a non sucky SPA where the users cant tell its a SPA. so you just get the pure advantages of partial page updates, async form validation, getting to use react for both partial & full page updates & server side rendering with shared code (rather than some janky combo of jquery and a server side template lang), etc. ajax was supposed to (and still can) make the web a more responsive ui. its just that spa frameworks dont come out of the box forcing you to do the right thing, so """web""" """"""developers"""""""" half rear end things like handling of all possible async states, proper url routing that doesnt break nav controls and linking, not making dynamic forms janky, etc. you have to come up with all this stuff on your own. web devs just seem more concerned with coming up with 1000 libraries to not have to type as much (stylus, jade, pug, coffeescript) Flat Daddy fucked around with this message at 18:00 on Oct 11, 2017 |
# ? Oct 11, 2017 17:56 |
|
i forked left-pad because i needed another parameter so now we pull the new left-pad from a local repo here
|
# ? Oct 11, 2017 17:59 |
|
Flat Daddy posted:its just that spa frameworks dont come out of the box forcing you to do the right thing, so """web""" """"""developers"""""""" half rear end things like handling of all possible async states, proper url routing that doesnt break nav controls and linking, not making dynamic forms janky, etc. you have to come up with all this stuff on your own. "batteries included" is ember's entire reason for existing and it's a shame it's not more popular.
|
# ? Oct 11, 2017 18:05 |
|
jony neuemonic posted:"batteries included" is ember's entire reason for existing and it's a shame it's not more popular. does it still have the problem where if you go 2 inches out of the happy path everything goes to hell e: serious question. last time i worked with that poo poo was in 2014
|
# ? Oct 11, 2017 18:08 |
|
bob dobbs is dead posted:does it still have the problem where if you go 2 inches out of the happy path everything goes to hell i haven't done anything complicated enough to go off the happy path tbh, but i suspect it's probably still not a great idea.
|
# ? Oct 11, 2017 18:18 |
|
quiggy posted:i get that it's valid c but it's loving obnoxious that g++ won't even throw a warning for that one. i had a construct like this: ah, my old nemesis code:
|
# ? Oct 11, 2017 18:43 |
|
LordSaturn posted:ah, my old nemesis
|
# ? Oct 11, 2017 18:51 |
quiggy posted:i get that it's valid c but it's loving obnoxious that g++ won't even throw a warning for that one. i had a construct like this: My g++ (7.2.0) will actually throw a hard error there: C++ code:
pre:$ g++ test.cpp -o test test.cpp: In function 'int main()': test.cpp:5:24: error: ISO C++ forbids comparison between pointer and integer [-fpermissive] std::cout << (p == '\0') << std::endl;
|
|
# ? Oct 11, 2017 21:24 |
|
VikingofRock posted:My g++ (7.2.0) will actually throw a hard error there: well mine doesn't, so there
|
# ? Oct 11, 2017 21:48 |
|
knockoutjs debugging guide: does it have enough parentheses? add more parentheses does it have too many parentheses? remove some parentheses parentheses
|
# ? Oct 11, 2017 21:58 |
|
it's 2017, why are you using knockoutjs?
|
# ? Oct 11, 2017 21:59 |
quiggy posted:well mine doesn't, so there update your software OP More seriously, does -Wconversion catch that on your g++? I would imagine it should, but I could see '\0' implicitly being given a pointer type to begin with and thus not being "converted".
|
|
# ? Oct 11, 2017 22:05 |
|
VikingofRock posted:update your software OP ah, here we go code:
|
# ? Oct 11, 2017 22:08 |
quiggy posted:ah, here we go goondolences I'm always shocked at how much C++11 improved things. I haven't looked into C++17 much yet, but hopefully it is a similar leap in quality.
|
|
# ? Oct 11, 2017 22:21 |
|
lol at the php-level gcc settings of Wall and Wextra
|
# ? Oct 11, 2017 23:26 |
|
on clang don’t forget -Weverything! (don’t use -Weverything)
|
# ? Oct 11, 2017 23:28 |
|
gcc -Wall -Wextra -pedantic --yes-i-want-all-the-warnings --if-its-a-warning-then-loving-enable-it -o horseshit garbage.c
|
# ? Oct 11, 2017 23:35 |
|
akadajet posted:it's 2017, why are you using knockoutjs? knockoutjs is the least bad client side rendering library
|
# ? Oct 11, 2017 23:38 |
|
knockoutjs is much worse than react+typescript but it's alright for small stuff. although I'd like to know if anyone is doing knockout without having to embed unchecked javascript code into html attributes because that's the only way I've really seen it done.
|
# ? Oct 12, 2017 00:25 |
|
imagine being a javascript graybeard. who will be the RMS of web front end bullshit?
|
# ? Oct 12, 2017 00:43 |
JewKiller 3000 posted:gcc -Wall -Wextra -pedantic --yes-i-want-all-the-warnings --if-its-a-warning-then-loving-enable-it -o horseshit garbage.c More like gcc -Wall -Wextra -pedantic --yes-i-want-all-the-warnings --if-its-a-warning-then-loving-enable-it --okay-actually-i-do-not-care-about-implicit-padding --nor-do-i-care-about-iso-c-compatibility --nor-double-promotion --nor-y2k-formatting --and-strict-overflow=5-is-completely-ridiculous-for-most-use-cases -o horseshit garbage.c VikingofRock fucked around with this message at 00:56 on Oct 12, 2017 |
|
# ? Oct 12, 2017 00:43 |
|
VikingofRock posted:goondolences not as big but a huge usability improvement. template parameter deduction has been finally extended to classes, poo poo like make_tuple or make_pair is no longer necessary. tuple expansion is native, no longer a library function. you could already map variadic template argument lists, now you can also reduce with any binary operator. the stl finally has long overdue essentials like optional, any, variant, string_view, gcd/lcm and the filesystem library. parallelized variants of algorithms
|
# ? Oct 12, 2017 00:57 |
|
Sapozhnik posted:some wunderkind on the c++ committee thought that "ptr == 0" looked more aesthetically pleasing than "ptr == NULL" lol, no they didnt. its because int* x = NULL doesn't work when you #define NULL (void*)0 in C++, but does in C because you can assign any void* implicitly in C and in C++ you need to explicitly cast it.
|
# ? Oct 12, 2017 01:00 |
|
akadajet posted:IE11 is super out of date wrt web standards and you know it. the enterprise doesn't give a poo poo about web standards, though. they want to continue to run their dumb Java 6 applets and ActiveX garbage until the CTO dies of old age, and they're used to just telling clients not to use other browsers also there are a bunch of weird edge cases that Edge doesn't handle properly, and reporting bugs to MS is an order of magnitude more difficult than with the other major browsers
|
# ? Oct 12, 2017 01:01 |
|
I wonder why it took them so long to come up with nullptr. if you wrote a smart pointer class and you wanted to make NULL assignable to it, guess how you had to do it?C++ code:
|
# ? Oct 12, 2017 01:03 |
hackbunny posted:not as big but a huge usability improvement. template parameter deduction has been finally extended to classes, poo poo like make_tuple or make_pair is no longer necessary. tuple expansion is native, no longer a library function. you could already map variadic template argument lists, now you can also reduce with any binary operator. the stl finally has long overdue essentials like optional, any, variant, string_view, gcd/lcm and the filesystem library. parallelized variants of algorithms Sounds pretty good to me! Very cool.
|
|
# ? Oct 12, 2017 01:05 |
|
whereas nullptr is a singleton, and you can explicitly match for its unique type. what took them so long? they already used very similar singletons for the non-throwing new operator and tag based dispatch
|
# ? Oct 12, 2017 01:07 |
|
CRIP EATIN BREAD posted:imagine being a javascript graybeard. in JS land, you're a graybeard for using a 2 years old framework
|
# ? Oct 12, 2017 01:12 |
|
CRIP EATIN BREAD posted:imagine being a javascript graybeard. not a graybeard, a graystache (probably waxed)
|
# ? Oct 12, 2017 01:16 |
|
i am once again rebuilding opencv from source because i forgot to include the path to the 3rd-party modules last time 2 months ago, whoops!
|
# ? Oct 12, 2017 01:20 |
Okay I'm reading through the C++17 changes and a bunch of these are very neat. My favorite so far (that hackbunny didn't mention) is attributes, which I had previously only run into in Rust. Being able to mark intentional switch-statement fallthrough as [[fallthrough]] (to suppress warnings when this is intentional) and being able to mark functions as [[nodiscard]] (to ensure that their return is used) are excellent.
|
|
# ? Oct 12, 2017 01:26 |
|
standard attribute syntax isn't new, c++11 already had [[noreturn]]
|
# ? Oct 12, 2017 01:31 |
|
oh right! another new c++17 feature is if constexpr, basically type-safe #ifdef. previously only achievable with gross template contortions
|
# ? Oct 12, 2017 01:36 |
hackbunny posted:standard attribute syntax isn't new, c++11 already had [[noreturn]] Huh, somehow I had missed that.
|
|
# ? Oct 12, 2017 01:36 |
|
and __has_include is now standard in the preprocessor! here's the full list: http://en.cppreference.com/w/cpp/compiler_support
|
# ? Oct 12, 2017 01:39 |
|
long as we're dorking out on c++ what are people's thoughts on vcpkg? i set it up last night to grab sdl and catch, seems to do what it says on the tin well enough.
|
# ? Oct 12, 2017 01:47 |
|
|
# ? May 15, 2024 15:53 |
|
jony neuemonic posted:long as we're dorking out on c++ what are people's thoughts on vcpkg? i set it up last night to grab sdl and catch, seems to do what it says on the tin well enough. it's fine, I guess, if you're only using msvc hackbunny posted:whereas nullptr is a singleton, and you can explicitly match for its unique type. what took them so long? they already used very similar singletons for the non-throwing new operator and tag based dispatch hell if know. remember it was standardized before 9/11 and also when FRIENDS was in its 4th season. it was a different time. Slurps Mad Rips fucked around with this message at 02:05 on Oct 12, 2017 |
# ? Oct 12, 2017 02:03 |