Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [tcf-dev] Context-query tid parameter issue 
2.) Update each comparator and inject the logic to handle hex and decimal values for each.
To use again your example with Linux PID, 2/ is stated under the assumption that it is valid for the user to enter an hexadecimal value for the PID when everywhere else the PID is shown as a decimal. Even worse, the service implementing the comparator may actually see the PID as a string and not a number and therefore doesn't event convert it to a number but instead uses strncmp().
As long as the properties type can't be requested by the client from the agent, I'm afraid you will have to keep some formatting rules for each property value in the client. 

- Christophe



My previous suggestion/request just seemed simpler.

Thanks -
Scott

-----Original Message-----
From: tcf-dev-bounces@xxxxxxxxxxx [mailto:tcf-dev-bounces@xxxxxxxxxxx] On Behalf Of Christophe Augier
Sent: Friday, September 21, 2012 9:06 AM
To: tcf-dev@xxxxxxxxxxx
Subject: Re: [tcf-dev] Context-query tid parameter issue

Hi Scott,

why don't you simply use double quotes all the time around your property value? From the service point of view, the property is either a number, a boolean or a string. This is used for parsing the context query but in the end the context query comparator function receives a string. So this is up to the service implementing the comparator to parse the property value.

- Christophe

On 9/21/2012 5:42 PM, Tepavich, Scott wrote:

When originally adding context query support for breakpoints, we only
tested against Linux.  Part of the query allows for the user to enter
a TID or PID for context filtering of the underlying OS attribute.
This worked perfectly due to the nature that Linux supports decimal
formatting of these fields.

Recently we attempted to test this same workflow for another RTOS.
This OS supports hexadecimal TID and PID values, and this form is more
natural for the user in this environment.  When these values are
placed in the context-query string and sent to the agent, the
parse_context_query() routine down in the agent returns an error
(since it's expecting a decimal value).

The interface defined for setting the context query, attempts to allow
the user to compose/edit the string by hand.  If we attempt to solve
the problem in the client, we would need to parse the string when the
user has completed entering their text, and convert the hex into decimal.
Either that or return an error to the user stating that hexadecimal is
not a valid form.

The latter is not a solution.  The former is problematic when
re-fetching the defined query for the user.  The values in the string
would be stored as decimal values, and the client would not have
enough criteria to choose which form to display to the user.  A "view
as hex | decimal" option could be inserted, but this can be
problematic where there are multiple numeric fields and each is
typically viewed in different forms.

This could be solved more simply, if we allowed a freeform expression
of numerical values as (at a minimum) decimal or hexadecimal for
context queries.  I pose the question if it would be acceptable if I
updated the context query parser to allow for this?  I think I already
know the answer, but feel compelled to ask...

Thanks for your time -

Scott



_______________________________________________
tcf-dev mailing list
tcf-dev@xxxxxxxxxxx
http://dev.eclipse.org/mailman/listinfo/tcf-dev


_______________________________________________
tcf-dev mailing list
tcf-dev@xxxxxxxxxxx
http://dev.eclipse.org/mailman/listinfo/tcf-dev
_______________________________________________
tcf-dev mailing list
tcf-dev@xxxxxxxxxxx
http://dev.eclipse.org/mailman/listinfo/tcf-dev

Back to the top