Add the user settable separator string for arrayed stats, default is standard ::

Information
Submitter:	brad danofsky
Repository:	gem5
Branch:
Bugs:
Depends On:
Reviewers
Groups:	Default
People:	ali, gblack, nate, stever

Description

Add the user settable separator string for arrayed stats, default is standard ::

I updated the diff to fix Gabes issues and I learned how to spell.

Testing Done

standard full regression

Why do you need this functionality? Why isn't "::" sufficient? If it's because some other piece of software is expecting something else, I'd say either write a script that munges these stats into a more acceptable format, or adjust the other software. I'm not an expert on our stats stuff by any stretch, but I'd hate to see extra otherwise unnecessary functionality be added to deal with a point issue that only affects a few people.

brad danofsky March 29, 2011, 2:11 a.m. (March 29, 2011, 2:11 a.m.)

You are correct that this may just be my problem.  I have a whole lot of scripts that expect a different character (one used in a previous version of m5 I think).  If it is objectionable that there be support for this then so be it.

Steve Reinhardt March 29, 2011, 8:16 a.m. (March 29, 2011, 8:16 a.m.)

These are scripts that date back to when we used a different separator (for SimpleScalar compatibility I believe), so arguably it's just fixing something we broke in m5 a long time ago.  I discussed this with Brad, and we decided that in the long run the ideal thing would be to leverage the any-day-now stats-in-python functionality to have m5 spit out stats directly in the format that these scripts currently generate, making the scripts obsolete.  But in the meantime it's a much smaller change to tweak m5 than to edit this pile of scripts, and I don't think it's hurting anyone to put this in there.

Gabe Black March 29, 2011, 9:03 a.m. (March 29, 2011, 9:03 a.m.)

It is a smaller change, but it's a smaller change that impacts -everybody- (at least theoretically) and only benefits a few people. But it's not a very big change and it sounds like it's temporary anyway (or at least that's the plan) so it's probably ok.

src/base/statistics.hh (Diff revision 1)

The bracket should be on the next line for functions, and should have a space before it if it were to go on that line.

brad danofsky March 29, 2011, 2:14 a.m. (March 29, 2011, 2:14 a.m.)

do you have an emacs mode hook for setting up your general style forms?

Gabe Black March 29, 2011, 2:21 a.m. (March 29, 2011, 2:21 a.m.)

I use vim so I don't use one myself, but util/emacs/m5-c-style.el looks like what you want.

src/base/statistics.hh (Diff revision 1)

This line is too long. You should spread it out like a normal function.

Summary:

-	Add the user settable seperator string for arrayed stats, default is standard ::
+	Add the user settable separator string for arrayed stats, default is standard ::

Description:

~		Add the user settable seperator string for arrayed stats, default is standard ::
	~	Add the user settable separator string for arrayed stats, default is standard ::

		I followed the flow for setting the description, name, etc. adding the ability to set the seperator string used
		between the array element and the name for vectors, vector2d, etc.
		One difference, that might be objectionable, is I made the stored string static.

Description:

		Add the user settable separator string for arrayed stats, default is standard ::
-
-		I followed the flow for setting the description, name, etc. adding the ability to set the seperator string used
-		between the array element and the name for vectors, vector2d, etc.
-		One difference, that might be objectionable, is I made the stored string static.