[PATCH] D33996: Improve consistency and flexibility with llvm-pdbdump subcommands

[Zachary Turner via Phabricator via llvm-commits]

zturner added inline comments.


================
Comment at: llvm/test/DebugInfo/PDB/pdbdump-readwrite.test:1
-RUN: llvm-pdbdump pdb2yaml -dbi-module-info -dbi-module-source-info \
-RUN:   -dbi-stream -pdb-stream -string-table -tpi-stream -stream-directory \
+RUN: llvm-pdbdump pdb2yaml -modules -module-files -dbi-stream \
+RUN:   -pdb-stream -string-table -tpi-stream -stream-directory \
----------------
inglorion wrote:
> While thinking about this: What does -dbi-stream mean? Is it useful to specify -dbi-stream without any of the additional enabling flags like -modules?
> 
> Also, should we at some point change this test to use -all instead of the various individual options?
If you specify `-dbi-stream` with nothing else it just dumps the header fields.  Perhaps a better way to organize all these options is to make them all by enum lists like I did with `-subsections`, then you could write: `-dbi=headers,modules,files,etc`.  Another option would be to make it so we don't do the hierarchical display at all, but instead make each option independent.  Then if you specified `-dbi-headers -modules -files` you'd get something like:

```
DBI Headers
--------------
...

Modules
--------------
...

Files
--------------
```

It's a little bit easier to do things from this method, but it's a little more separated from the underlying file structure which is kind of disadvantageous.


https://reviews.llvm.org/D33996