Thats a very nice find, thanks for the link. I'm not entirely sure about putting everything into a header file -- and marking load of pretty big functions as 'inline' but well, it simplifies the build a lot it seems!
What I appreciate most is that he is upfront and honest about it so nobody needs waste time making changes that won't be accepted. (Or if they do, they know that they're forking). As an approach it is understandable for something that solves your problem but which you have no intention of maintaining but it won't help a project be successful.
Most of the benefit of open source projects is that they are open to collaboration, I would avoid even using a project with policies like that if I knew about them.
That seems to be in line with what the author wants. Plenty of people prefer open source to mean "you can copy my code," and not "let's build something together". You using, or not using, the author's code will likely have no impact on him.
I have a variety of open source projects, and some fall towards each end of the spectrum.
understandable, athough I can also understand his position.
Actually I think he himself just wants to not collaborate, but I don't see anything saying that you can't copy his code and then collaborate with others.
There are also other benefits of open source other than collaboration, such as security from being able to review the code, and being able to run in a totally free system.
A toolkit like this is only another option in your basket if you can ignore real-world considerations like:
- Internationalization: The edit control doesn't support input method editors. On Windows, wide-character functions aren't used; that means Unicode isn't supported.
- Accessibility: Here I'm talking about the ability for users with disabilities, most notably blind users but also users with mobility impairments, to access a UI via assistive technologies such as screen readers or alternative types of input. This requires the GUI toolkit to implement an accessibility API such as Microsoft Active Accessibility or UI Automation on Windows, or AT-SPI for X11-based environments. This is something that practically everyone who sets out to implement a lean, mean, simple toolkit just doesn't pay attention to.
- Native look: The toolkit doesn't attempt to mimic the look of the rest of the system at all. On my Windows 10 machine, the demo looks like a throwback to Windows 95. This is also an accessibility issue, for low-vision users that need a different color scheme.
And there are probably other considerations I've forgotten or am unaware of. You'd probably be better off using one of those big, bloated C++ toolkits, like Qt or (better yet, IMO) wxWidgets. Not GTK, since that doesn't cover accessibility except under X11 via GNOME.
Looks like it supports X and W32. Unless it's using a lot of the crufty "draw a stippled line" sort of functionality that X provides it would be pretty easy to port. That is, if all it does is construct a bitmap in memory and hand it off to the compositor, it's practically Wayland-ready.
- It used to be that Qt was massive; because it existed before ubiquitous, usable STL implementations, Qt had to essentially also ship its own C++ standard library. Which meant that you couldn't use standard C++ strings, you had to use QString, and so on. This is less true with Qt 4 & 5, but reputations stick around.
- Often-times C++ APIs are a pain, even if you are using C++, many people have the opinion that the API exposed should be C.
STL strings are also next to useless. startsWith(), endsWith(), split(), case-insensitive find, stringify a number, numericify a string, easily construct a formated string, properly handles unicode (wstring does not), all require extra code. QString has a function for each of those.
And the STL data structures are sometimes kind of unwieldy:
if (dict.find(key) != dict.end()) { ... }
is lame compared to the easily understood:
if (dict.contains(key)) { ... }
Not to mention all the stupid inconsistencies:
// std::vector has no sort function, haha!
std::sort(std_vector.begin(), std_vector.end());
// std::sort doesn't work on std::list, haha!
std_list.sort();
And my favorite annoyance, the fact that std::vector::size() returns unsigned int, despite the fact that for practical purposes you aren't going to get anywhere near even 2 billion elements without running out of memory (or other performance problems), so I'm stuck writing
for (unsigned int idx = 0; idx < v.size(); ++idx) {...}
or omitting the "unsigned" and getting lots of compiler warnings. I get why they would have done it; I would have done it too. Until I used Qt, then I like how everything is simply int. I suppose if you are doing physics simulations or something you might need something different, but it's the rare GUI program that needs unsigned int for size.
vector::size() actually returns size_t, which is exactly the right way to go about, because why should a size() function be able to ever return negative numbers?
Good point about size_t, although technically it's std::size_t.
I agree, STL has done the "right" thing with sizes; of course negative sizes are meaningless. The problem is, Qt's way is so much nicer to use, even though it is obviously "wrong." Type three characters, done. Putting "int" in your headers, it's obvious what's going on--you're getting a number back. I suppose std::size_t is obvious, but it's hard to read, and I hate typing underscores. Again, more correct, but I hate doing it, which is just another reason why Qt is such fun to use compare to STL.
Yeah, I know why. Problem is, it is a bad design. I don't work for the API, the API works for me. Conceptually sorting a list is the same as sorting a vector, so it should look the same in the code. The fact that it may require a different algorithm underneath is what the API is there to figure out for me. Other languages and APIs (including Qt) don't have problems like this, hence my complaint.
You're unlikely to have a std::vector containing 2 billion single byte elements however.
Even being conservative, say you had a struct of size 24 bytes (e.g. 3 doubles x, y and z) then now you're up to 48G.
So yes, there are very few situations where you'll have meaningful data that would fill a vector indexed by signed ints without running out of memory first.
Can you imagine how stupid it would feel if you had to abandon all your std::vector-using code the moment you need to deal with an array bigger than 2 billion elements?
Actually, thinking about it a bit more, the only thing you could store in a std::vector indexed by signed ints where you would have a problem would be single bytes.
If you have 2 billion 2 byte elements then that's 4gb which is the total addressable space on a 32-bit system - leaving no room for program code or any other data meaning you couldn't run anything.
You could go 64bit, but then signed ints go up to (2^63)-1 and you have the same problem - elements more than a single byte in size will cause you to run out of addressable memory before you exhaust available signed ints.
Yes, and like a vector containing > 2 billion single byte elements it's still delving in to a very peculiar use case.
Compared to the majority of other uses of std::vector which will never be able to overflow signed ints.
I get from a logical point of view that size() would never be < 0 therefore it makes sense to use an unsigned int, however from a practical point of view it also makes programs more susceptible to a number of pernicious bugs that can catch out unsuspecting programmers, e.g. things like
for ( size_t i = v.size(); i >= 0; --i )
{
...
}
Which would work fine with a signed int, but here it will just loop infinitely.
- I'm in the camp that thinks widget libraries for C++ should be workable in C.
- I'm big on minimalism, and being strongly aware of "90% of programs out there are bloated", and was horrified when I had to install a tiny Qt based program in ubuntu, and it required me to install a gazillion packages that looked like they belonged to Qt. (I don't remember the details).
- I always keep looking for C programs that can do in a couple thousand lines, what takes tens of thousands of non-C code for other programs to accomplish. I'm fairly aligned with the suckless philosophy, although I may not 100% agree with them. (suckless website would give you a good idea of what I'm talking about).
Also not the op, but I personally tend to avoid Qt if I can help it because I prefer to release software under MIT/Apache/BSD style licenses, which is incompatible with GPL style licenses (in that direction), or makes things more complicated at the very least.
As long as you don't use anything derived from QObject, but that pretty much defeats the purpose...
The moc is to effectively extend C++ with signals and slot at a language level. I'm not crazy about the moc, but I love signals and slots. The only alternative is slow-compiling templates like boost signals and slots, which also have less than compelling syntax.
Another lightweight ANSI C library is https://github.com/vurtun/zahnrad. It is platform and render backend independent, has no dependencies and is configurable.
I wanted something like this and was reading about GUI toolkits when I decided to take a moment to check HN. Talk about good timing!
On the other hand... I kind of want vector graphics, and cairo would be really great for that, and if I use cairo, well I might as well use gtk, and... hmm. I see why these things end up with lots of features.
C syntax is simple and elegant. Same can be said to Go and Lua, the later two have additionally very good inbuilt string support. The BNF of these languages fits on one page
wow, I dreamed I was in 2015 with powerful pocket size computers showing flashing GUIs with multitouch support.
Darn.. I just woke up, I'm back in the 80's
It works with NanoVG, so it is useful in environments where you're already running OpenGL.