Re: [PATCH 3/8] Change fde table to a vector

[Tom Tromey <tom@tromey.com>]

>>>>> "Simon" == Simon Marchi <simark@simark.ca> writes:

Simon> As a further change, I think we could consider make this a vector
Simon> of objects, rather than a vector of pointers.

I think I'd rather leave this for a future change.
It makes the patch a lot less obvious, to me anyway.  find_fde returns a
pointer -- but is that safe?  I think so, but I am not 100% certain; but
with the current code it definitely is.

>> +const struct objfile_key<dwarf2_fde_table> dwarf2_frame_objfile_data;

Simon> Can you make this static, while at it?

I did this.

>> +  if (!fde_table.empty ())

Simon> If you want to simplify the code even further, you could get rid
Simon> of this check, since the code below should cope well with an
Simon> empty vector.

Thanks, I did this too.

>> +      fde_table2->shrink_to_fit ();

Simon> That shrink_to_fit appears to be pointless, as fde_table2 is
Simon> created empty and we only push_back.  Maybe you meant to create
Simon> it with an initial size?

IIUC push_back normally uses some resize factor; the shrink_to_fit is a
request to discard any excess memory allocated this way.

FWIW another thing that could be done in this code is to use the
remove_if to remove elements in-place from fde_table.  This is another
thing I'd rather do separately.

Tom