Minor discrepancy between SetOffset() and WriteTo() about dnlib HOT 5 CLOSED

Are you changing heap.Name between calls to SetOffset() and WriteTo()? That will break the code since obviously SetOffset() must know the name in order to calculate its size.

if not, do you have a sample that triggers this condition?

from dnlib.

For example, if the reserved Flags value (ECMA-335 name) is not written, SetOffset expects 108 bytes to be written but only 106 bytes are actually written to the output although the code changes are valid:

SetOffset(...) {
....
//length += 4;   // Flags and Streams
length += 2;   // Streams only
...

;

WriteTo(...) {
...
//writer.Write((byte)(options.StorageFlags ?? 0));
//writer.Write(options.Reserved2 ?? 0);

I understand that almost noone modifies the header this way and I expect this behaviour to be resolved as "Won't Fix"; I just wanted to let you know.

from dnlib.

You've modified the writer code so instead of writing StorageFlags,Reserved2,heaps.Count, you now only write heaps.Count? How does the CLR know this and correctly parses the new header without the 2 bytes before the heap count field? Is there a new flag I don't know about or forgot?

from dnlib.

The CLR can't parse it, I work on my own runtime. I should have mentioned it before.

from dnlib.

This header will always be 4-byte aligned so there's only a problem if you change the size of the header so the heaps aren't 4-byte aligned, as you did in your modified code. I'll close this as a Won't-Fix 😎 but it should be trivial for you to fix it for your custom header.

from dnlib.