stdcxx-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Anton Pevtsov <>
Subject Re: Re: test for 21.string::erase
Date Tue, 14 Mar 2006 16:31:13 GMT
I've added command line options: --no-exceptions, --no-erase,
--no-erase-iterator, --no-erase-range, --no-erase-pos, and
The attached file contains the patch for the test.

Here is the ChangeLog: 

2006-03-14  Anton Pevtsov  <>

	* 21.string.erase.cpp: (rw_opt_no_exceptions, rw_opt_no_erase, 
	rw_opt_no_erase_pos, rw_opt_no_erase_count,
	rw_opt_no_erase_range) : New command line options.
	(test_erase) : Added check for the --no-exceptions option.
	(note_test_disabled) : New function to display the disable note.
	(run_test) : Changed to support new command line options.
	(main) : Added new command line options.

Anton Pevtsov

-----Original Message-----
From: Martin Sebor [] 
Sent: Sunday, March 12, 2006 04:08
Subject: Re: test for 21.string::erase

Anton Pevtsov wrote:

>> The attached file contains new version of the test for string::erase. 
>> I merged tests for all erase versions into one function and added 
>> UserChar's tests.

Okay, thanks. I found a few small problems and decided to correct them
myself and commit the result -- see:

Attached is a diff of the changes. The most important one is to the
CALLFMT and CALLARGS macros where the test would fail to print the
string argument past the first NUL. The other (minor) issues were the
bad spelling of the $Id$ RCS keyword (missing closing $) and the file
name (should be 21.string.erase). Please look over the diff so that we
can avoid running into the same issues in the future  :) 

Also, I would like each test to be able to disable exceptions using a
command line option (say, "--no-exceptions"). Could you please add this
capability to this test? Finally, it would be good to provide options to
disable each overload of the tested function, i.e., --no-erase,
--no-erase-iterator, --no-erase-range, --no-erase-pos, and
no-erase-count. Let's also use this naming convention in the rest of the
string and container tests. We should also have a uniform naming
convention for the options that control the user defined character type
and user-defined traits (I used --no-UserTraits in 0.char.cpp but I am
willing to change it to --no-user_traits of you like that better).

Thanks again.

View raw message