20% space savings!

20% space savings!

[Michael Elizabeth Chastain]

I pulled a Daniel Berlin last night (in a good way) and started
packing structures like a #4 train at rush hour.

'maint space' has gone from:

  before: 99352576
  after:  79478784

It would take a lot of time to merge these patches into gdb.
I would have to write up a road map, and then test the patches
one at a time (that is the hard part, regression testing stuff
like coff), then submit the patches one at a time to the
symtab maintainers.

Is this something worth pursuing?

Michael C

Re: 20% space savings!

[Kevin Buettner]

On Aug 19,  2:14pm, Michael Elizabeth Chastain wrote:

> I pulled a Daniel Berlin last night (in a good way) and started
> packing structures like a #4 train at rush hour.
> 
> 'maint space' has gone from:
> 
>   before: 99352576
>   after:  79478784
> 
> It would take a lot of time to merge these patches into gdb.
> I would have to write up a road map, and then test the patches
> one at a time (that is the hard part, regression testing stuff
> like coff), then submit the patches one at a time to the
> symtab maintainers.
> 
> Is this something worth pursuing?

I think it's worth pursuing.  For large applications, a 20% savings in
space could mean the difference between being able to easily debug the
app and running out of memory (or running into swap).

Kevin

Re: 20% space savings!

[Daniel Jacobowitz]

On Tue, Aug 19, 2003 at 02:14:29PM -0400, Michael Elizabeth Chastain wrote:
> I pulled a Daniel Berlin last night (in a good way) and started
> packing structures like a #4 train at rush hour.
> 
> 'maint space' has gone from:
> 
>   before: 99352576
>   after:  79478784
> 
> It would take a lot of time to merge these patches into gdb.
> I would have to write up a road map, and then test the patches
> one at a time (that is the hard part, regression testing stuff
> like coff), then submit the patches one at a time to the
> symtab maintainers.
> 
> Is this something worth pursuing?

Yes.

I think so, at least.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: 20% space savings!

[Andrew Cagney]

> I pulled a Daniel Berlin last night (in a good way) and started
> packing structures like a #4 train at rush hour.
> 
> 'maint space' has gone from:
> 
>   before: 99352576
>   after:  79478784
> 
> It would take a lot of time to merge these patches into gdb.
> I would have to write up a road map, and then test the patches
> one at a time (that is the hard part, regression testing stuff
> like coff), then submit the patches one at a time to the
> symtab maintainers.
> 
> Is this something worth pursuing?

This will definitly help short term.  Long term, to abuse an ad, "While 
amazing, it's not sensational[sp?]".  As David and Elena work through an 
overhaul of the symbol table and a better understanding of what it 
really needs to do is gained, a more radical approach can be taken.

Andrew

User defined commands

[jacques]

Hello,
I've been searching for the easiest implementation to be able to inspect 
stl containers, and I've run into an intriguing possibility, using user 
defined commands. I was wondering if its possible to put a statement "if 
a is vector" in gdb?