Hans Wennborg

I like to program computers. @[email protected]

[Hans Wennborg @hansw2000]

New blog post! Shrink, Reduce, and Implode:
The Legacy Zip Compression Methods
https://t.co/GafOWq9lMu

[Hans Wennborg @hansw2000]

@smarr Not exactly what you're asking for, but https://t.co/O2slU5AODm may be a good starting point.

[Hans Wennborg @hansw2000]

@TalBeerySec @aionescu @markrussinovich @lcamtuf @ericlaw @zodiacon https://t.co/OYWyhb0RRX might be a good start

[Hans Wennborg @hansw2000]

@colinianking What was #1?

[llvm.org @llvmorg]

We are pleased to announce the 2024 LLVM Developers' Meeting program! Register today! https://t.co/6g1bRSrKxy

[Hans Wennborg @hansw2000]

@angealbertini The MS-DOS Encyclopedia just refers to them as ".EXE files" https://t.co/XDMenkzZdV

[Hans Wennborg @hansw2000]

@galinash The wording makes it sound scammy

[Hans Wennborg @hansw2000]

@rui314 New desk setup? :)

[Hans Wennborg @hansw2000]

@ericlaw @ProgramMax Yup. I’m guessing that there’s room for improvement though :)

[Hans Wennborg @hansw2000]

Michael Kerrisk's The Linux Programming Interface (https://t.co/rFDGOXdUU1) looks amazing. Is there anything equivalent for Win32?

[Hans Wennborg @hansw2000]

@angealbertini @mer0x36 @greatcodeholio https://t.co/Tl4TXVDEkM mentions example "ZM" binaries from IBM (presumably created by that linker?), and note that originally https://t.co/NMrbi7dhXM did not check the magic (https://t.co/VHZiJAetcs seems to check the extension only), which is presumably how this could happen.

[Hans Wennborg @hansw2000]

@angealbertini @mer0x36 @greatcodeholio Mark told me: "ZM was emitted by the linker IBM wrote for their mainframe in '81. For some reason big-/little-endian was confusing for them."

[Hans Wennborg @hansw2000]

@fsfodx @JokerEph Yup, that's the one.

[Hans Wennborg @hansw2000]

Savage 😄

[Hans Wennborg @hansw2000]

@sasank51 The slide is from "A Wishlist for Faster LLVM Back-ends" in this session: https://t.co/LjY2o8sQHo I don't think the author is on Twitter. They presented a good analysis of things slowing down the -O0 pipeline. The "should we start over" remark referred to https://t.co/f6FfuK9xpk

[Filip Jerzy Pizło @filpizlo]

In my spare time, I made a memory-safe C that I call Fil-C. https://t.co/P00qxQ2xI0

[Hans Wennborg @hansw2000]

The craziest build issue I've debugged in a long time: https://t.co/Z2nZObWko9

[Hans Wennborg @hansw2000]

https://t.co/214QQLGMQ8 (and the whole thread) is a great example of how hard it is to reason about what is/isn't allowed by the C++ standard.

[Hans Wennborg @hansw2000]

@moyix Maybe @SylvestreLedru knows?

In case it's useful, Chromium keeps its toolchain packages here: https://t.co/UdxSaAOu9M

[Nikita Popov @nikita_ppv]

New blog post: This year in LLVM (2023)
https://t.co/vfH0AAJyHL

[Hans Wennborg @hansw2000]

@rsms I believe the language doesn't allow it: https://t.co/v5gAKipcmL

As Rui mentions, ICF in the linker could perhaps do this, but currently that only works with section granularity.