develooper Front page | perl.perl5.porters | Postings from January 2008

Re: dtrace and Perl (again)

Thread Previous | Thread Next
From:
Nicholas Clark
Date:
January 7, 2008 15:30
Subject:
Re: dtrace and Perl (again)
Message ID:
20080107233049.GD23703@plum.flirble.org
On Mon, Jan 07, 2008 at 11:21:22PM +0000, Andy Armstrong wrote:
> On 7 Jan 2008, at 23:15, Nicholas Clark wrote:
> >>I assume that all the above should be conditional on -Dusedtrace and
> >>that Apple and Sun will set that option when they build their vendor
> >>versions. Is that sensible?
> >
> >No! :-)
> >
> >Only in that if it's the same speed, I don't see why we don't  
> >default to
> >building it in on any platform where we can, and let the user have  
> >to do
> >-Uusedtrace if they don't want it.
> 
> I guess I'm worried about introducing fragility into the build. There  
> seem to be a few different versions of dtrace floating around - some  
> of which are broken. There's at least one extant Sun version that  
> can't do the .d -> .h transformation for example.
> 
> My thinking is that if it's on by default then broken dtrace implies  
> broken Perl.

Good point.

Default to on if -Dusedevel is present and dtrace is found, and off otherwise?

> That's the other thing that attracted me to an explicit switch. Isn't  
> asking for dtrace support and having the build fail because it's not  
> available better than hoping for dtrace support and having it omitted  
> because we couldn't find the executable?

Is it viable to probe for a working dtrace? Do we know what is sufficient
to make a test case that the broken ones reliably choke on?

[Clearly get it in manually first, then worry about this. Plus Merijn is busy
with back porting currently so large Configure patches aren't a great idea
right now]

Nicholas Clark

Thread Previous | Thread Next


nntp.perl.org: Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at ask@perl.org | Group listing | About