Long preloaded file not properly read on Windows

[hjenryin]

Version of emscripten/emsdk:

emcc (Emscripten gcc/clang-like replacement + linker emulating GNU ld) 3.1.42 (6ede0b8fc1c979bb206148804bfb48b472ccc3da)
clang version 17.0.0 (https://github.com/llvm/llvm-project f3b64887de61020c09404bfee97b2fadd30df10a)
Target: wasm32-unknown-emscripten
Thread model: posix

Bug Description

In short, if a very long file (in my case, 116363 lines) is preloaded using --preload-file, nothing will go wrong when opening the html or js on linux. However on Windows, not all lines are included (in my case, only 112130 lines could be read), no matter using node or a browser.

A reproducing example can be found at https://github.com/hjenryin/wasm_test. How results can be reproduced can be found in the readme file.

[sbc100]

I can't seem to replicate on my windows machine. I get the same on windows as on linux:

C:\Users\circleci\project\wasm_test>%EMSDK_NODE% main.js
File at data.json
Number of lines in file: 116362

Which version of node are you using on windows?

Is the size of the resulting wasm file the same on windows a linux for you? I see 170930 bytes on both platforms

[hjenryin]

I tried 18.15.0 and 18.17.0 for Node, and both result in

PS D:\Programming\wasm_test\build_js> node .\main.js
File at data.json
Number of lines in file: 112130

Since I'm using Visual Studio 17 2022 generator for cmake on my windows machine, the emmake wrapper doesn't quite work. So I'm just using the files built on linux directly (which are the ones in the repo). ls on windows gets

Mode                 LastWriteTime         Length Name
----                 -------------         ------ ----
-a----          2023/8/2     22:06        2970078 main.data
-a----          2023/8/2     22:06         103765 main.html
-a----          2023/8/2     22:06         200353 main.js
-a----          2023/8/2     22:06         170977 main.wasm

main.wasm is also of size 170977 on my linux machine. Other sizes are different due to different line break type.

[sbc100]

Can you perhaps attach the bad and the good wasm files so we can compare them?

[hjenryin]

Hi, sorry for the delay. I must have missed your reply.
I think you misunderstood me because I'm using the same wasm file all the time. For reproduction, you can take a look at https://github.com/hjenryin/wasm_test/actions/runs/7798941174 , where I run it on both Linux and Windows machines provided by Github. The results were same as my observation.