Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Pystd, similar-ish functionality with a fraction of the compile time

nibblestew.blogspot.com

Post reply on HN
Print view View original

21–30 of 32 posts

Re: Pystd, similar-ish functionality with a fraction of the compile time

#21

Post by integricho »

pjmlp wrote: post #20
integricho wrote: post #19

Earlier quoted context omitted.

Have you checked what dll dependencies you get if you build such an app? And have you actually tried running such an executable on win9x?

If you use only Win32 APIs from C++ just like since Petzold books, the dlls are already there, there is also static linking for other stuff. Win 9x is a dead OS, why would I bother with that outside retro computing?

Understood, yes, but I described my reasons in my reply to another comment already. Your proposed solution is not equivalent to what I am doing, it cannot work on all the systems I support, and would never be as dependency free.

Re: Pystd, similar-ish functionality with a fraction of the compile time

#22

Post by pjmlp »

integricho wrote: post #21
pjmlp wrote: post #20

Earlier quoted context omitted.

If you use only Win32 APIs from C++ just like since Petzold books, the dlls are already there, there is also static linking for other stuff. Win 9x is a dead OS, why would I bother with that outside retro computing?

Understood, yes, but I described my reasons in my reply to another comment already. Your proposed solution is not equivalent to what I am doing, it cannot work on all the systems I support, and would never be as dependency free.

I fail to see why, as I have been doing C++ on Windows since Windows 3.0.

I only don't consider legacy Windows platforms something to care about, and put the required effort into making something like that happen.

Re: Pystd, similar-ish functionality with a fraction of the compile time

#23

Post by integricho »

pjmlp wrote: post #22
integricho wrote: post #21

Earlier quoted context omitted.

Understood, yes, but I described my reasons in my reply to another comment already. Your proposed solution is not equivalent to what I am doing, it cannot work on all the systems I support, and would never be as dependency free.

I fail to see why, as I have been doing C++ on Windows since Windows 3.0. I only don't consider legacy Windows platforms something to care about, and put the required effort into making something like that happen.

Even trivial stuff causes the linking to vcruntime, this is an extra dll dependency that I don't tolerate, no msvcrt, no vcruntime, nothing except core Win32 dlls are allowed on the platform. Static linking can relieve some of this pain, but that bloats the binary. C++ simply does not allow the same level of minimalism that can be achieved by C, and for many these are unimportant details, for me they are deal breakers and this is a core pillar of my architecture.

Re: Pystd, similar-ish functionality with a fraction of the compile time

#26

Post by pjmlp »

integricho wrote: post #23
pjmlp wrote: post #22

Earlier quoted context omitted.

I fail to see why, as I have been doing C++ on Windows since Windows 3.0. I only don't consider legacy Windows platforms something to care about, and put the required effort into making something like that happen.

Even trivial stuff causes the linking to vcruntime, this is an extra dll dependency that I don't tolerate, no msvcrt, no vcruntime, nothing except core Win32 dlls are allowed on the platform. Static linking can relieve some of this pain, but that bloats the binary. C++ simply does not allow the same level of minimalism that can be achieved by C, and for many these are unimportant details, for me they are deal breaker…

First of all VC++ isn't the only C++ compiler on Windows, as it isn't like UNIX with the one true compiler original approach, secondly there are ways with the compiler and linker flags.

I have been avoiding C as much as possible since 1992, starting with Turbo C++ for MS-DOS.

Anyway, to each its own, all the best.

Re: Pystd, similar-ish functionality with a fraction of the compile time

#27

Post by ncruces »

olq_plo wrote: post #25

Dumb. This is what modules are for. Also, the stdlib is extremely well designed. It considers edge cases most people never think about. Source: I am a Boost Developer.

The first word in your comment spoils the rest of it.

Re: Pystd, similar-ish functionality with a fraction of the compile time

#28

Post by nok22kon »

pjmlp wrote: post #16
nok22kon wrote: post #13

Earlier quoted context omitted.

wasn't there a lot of talk that modules are still not really working, in practice I mean?

Depends on where you are. VC++ and clang latest with MSBuild or CMake/ninja are there, minus some bugs or code completion misbehaving (but bearable). GCC 16 is mostly ok now, also with CMake/ninja. All my hobby coding in C++ makes use of modules, at work it is a different matter, where libraries to be consumed by Java/.NET/nodejs, are still using C++17 as baseline. You can easily check, https://github.com/pjmlp/Raytr…

> at work it is a different matter

that's what I was asking, of modules are being introduced at scale or if they are a hobbist thing

https://nibblestew.blogspot.com/2025/08/we-need-to-seriously...

Re: Pystd, similar-ish functionality with a fraction of the compile time

#29

Post by integricho »

pjmlp wrote: post #26
integricho wrote: post #23

Earlier quoted context omitted.

Even trivial stuff causes the linking to vcruntime, this is an extra dll dependency that I don't tolerate, no msvcrt, no vcruntime, nothing except core Win32 dlls are allowed on the platform. Static linking can relieve some of this pain, but that bloats the binary. C++ simply does not allow the same level of minimalism that can be achieved by C, and for many these are unimportant details, for me they are deal breaker…

First of all VC++ isn't the only C++ compiler on Windows, as it isn't like UNIX with the one true compiler original approach, secondly there are ways with the compiler and linker flags. I have been avoiding C as much as possible since 1992, starting with Turbo C++ for MS-DOS. Anyway, to each its own, all the best.

Sure, just to clarify, I am not arguing that c++ doesn't work, only that c++ produced binaries are either having more dll dependencies, or are more bloated (in size). I also tried clang-cl and MingW, and while both work, the binaries produced by them are even larger then any MSVC produced (regardless of how I try to optimize it). C is a winner when it comes to binary size.

Re: Pystd, similar-ish functionality with a fraction of the compile time

#30

Post by pjmlp »

nok22kon wrote: post #28
pjmlp wrote: post #16

Earlier quoted context omitted.

Depends on where you are. VC++ and clang latest with MSBuild or CMake/ninja are there, minus some bugs or code completion misbehaving (but bearable). GCC 16 is mostly ok now, also with CMake/ninja. All my hobby coding in C++ makes use of modules, at work it is a different matter, where libraries to be consumed by Java/.NET/nodejs, are still using C++17 as baseline. You can easily check, https://github.com/pjmlp/Raytr…

> at work it is a different matter that's what I was asking, of modules are being introduced at scale or if they are a hobbist thing https://nibblestew.blogspot.com/2025/08/we-need-to-seriously...

Microsoft Office uses modules, so definitely not only hobbyist.
Post reply on HN