Brad Larsen

The `instructions` slice inside of `gix_pack::data::file::decode::entry::::resolve_deltas` is being computed differently on my ARM64 and x86_64 machine. Investigating further...

Hmmm, well, it does seem like something is coming out differently on ARM64 vs x86_64 from the call to `inflate.once()` in this: https://github.com/GitoxideLabs/gitoxide/blob/fce70950006892f51b32af233656be6fe5de9df3/gix-pack/src/data/file/decode/entry.rs#L119-L132. I haven't debugged further into the miniz_oxide...

> Can you get the input/output pair out for the misbehaving target? I'll try to grab this next time I look at this, probably next week

>> Wait, did you already post the "bad" input/output somewhere? That's the interesting one really. Or is there a project/branch I can run to replicate? I'm on x86_64 linux. >...

> I wonder why it works at all, and also why the produced file seems to be PKM 10@X - is that correct even? I presume nosey parker would have...

Note, here's the content that `git version 2.34.1` prints out for that blob: ``` % git show 5665283d8c34f78a615fb6bf76fdd14bdb19b882 | hexdump -C 00000000 50 4b 4d 20 31 30 00 00...

@Byron running with your debug output patch on top of commit c7af04db9b6bb1204e0f4c436d1db8f48a491e86 on my x86_64 machine, I see the [same input as you got on ARM64](https://github.com/GitoxideLabs/gitoxide/issues/2044#issuecomment-3030790666), but an output that...

> @bradlarsen Do I understand correctly that NoseyParker found that object by traversing the commit-graph exclusively? I am asking because in theory, any kind of corrupted pack could be sent,...

I'm still sorting out a hosting solution for these docs. In the meantime, here are offline-generated docs for the current snapshot: [site.tgz](https://github.com/user-attachments/files/18679684/site.tgz)