Re: [PATCH] Conditional tracepoints

[Stan Shebs <stan@codesourcery.com>]

Eli Zaretskii wrote:
>> Date: Fri, 26 Jun 2009 22:31:07 -0700
>> From: Stan Shebs <stan@codesourcery.com>
>>
>> This patch adds conditional tracepoints.
>>     
>
> Thanks.
>
> I think we need a NEWS entry for this.
>   
OK, good idea.
>> + programming language (@pxref{Expressions, ,Expressions}).  A
>> + tracepoint with a condition evaluates the expression each time your
>> + program reaches it, and data collection happens only if the condition
>> + is @emph{true}.
>>     
>
> Not sure why you use the @emph markup here.  I think the text is okay
> without any markup at all.
>   
It's been several months since I wrote it, and now I don't remember 
either. :-)
>   
>>   @table @code
>>   @kindex maint agent
>> + @kindex maint agent-eval
>>   @item maint agent @var{expression}
>> + @itemx maint agent-eval @var{expression}
>>   Translate the given @var{expression} into remote agent bytecodes.
>>   This command is useful for debugging the Agent Expression mechanism
>> ! (@pxref{Agent Expressions}).  The @samp{agent} version produces an
>> ! expression useful for data collection, such as by tracepoints, while
>> ! @samp{maint agent-eval} produces an expression that evaluates directly
>> ! to a result.
>>     
>
> I must admit that I don't understand the description you added.  What
> does "expression useful for data collection" mean, and how is it
> different from "expression that evaluates directly to a result"?
>   
Expressions compiled for tracing have bytecodes that cause saving of the 
blocks of memory etc that would be needed back on the host to evaluate 
the expression at some later date.  It occurs to me that the Agent 
Expressions section should say a little more about the two kind of 
expressions.

>> + @item ConditionalTracepoints
>> + The remote stub accepts and tests conditional expressions defined for
>> + tracepoints (@pxref{Tracepoint Conditions}).
>>     
>
> Don't you mean "accepts and supports" or "accepts and implements"?
>   
"support" and "implement" seemed overused. :-)

Stan