options to sort

[martin-paul]

Hi,

first of all thanks for the great tool - it helps investigating the DWARF structure.
It is neverthelss quite hard to navigate in a bigger dwarf file. A major improvement for navigating the DIEs would be the possibility to give the option to sort also by name instead of debug_info address (applicable for all hierarchy levels simulatnously)

[Seva Alekseyev (he/him)]

Please clarify the "all levels" thing. Do you mean - ignore the tree hierarchy and just list the DIEs with a name in sorted order?

Also, most DIEs don't have a name.

What's your business case, anyway? Are you having trouble finding a particular function or class? That's what sorting is for.

[martin-paul]

with all level sorting i mean sort within each hierarchy level:

if the current structure would look like this

test.cpp

• DW_TAG_namespace:std

main.cpp

• DW_TAG_namespace: std
  ** DW_TAG_imported_module
  ** DW_TAG_const_type
• DW_TAG_const_type

after sorting within all hierarchy levels would look like:

main.cpp

• DW_TAG_const_type
• DW_TAG_namespace: std
  ** DW_TAG_const_type
  ** DW_TAG_imported_module

test.cpp

• DW_TAG_namespace:std

I am currently writing a dwarf parser & extractor in java. Your tool helps me investigating bugs 👍 but when i check for a larger DWARF file and check for certain DIEs, it can take quite a while to naviagate to the element of interest (at later stages of my tool, the relative CU addresses are no longer accessible so i need to rely on for example on TAG + filename + namespace + name)

[Seva Alekseyev (he/him)]

I see what you mean. I misunderstood, thought you meant sorting by the name attribute, but it looks what you're after is sorting by DIE tag, then by name.

It's possible, give me a couple of days. I'll probably decouple sorting CUs in the file from sorting DIEs - there's an obvious, natural sorting order for CUs, and it makes sense to make them sorted by default.

That said, if sorted, a typical CU child set would have a bunch of array_types, then base_types, then const_types, and so forth. I hope that's what you're looking for :)

[martin-paul]

yes exactly, thanks in advance ! :)

[Seva Alekseyev (he/him)]

Published, please take a look.

[martin-paul]

works like a charm - thanks a lot sevaa ! :)