From: Pedro Alves <palves@redhat.com>
To: Tom Tromey <tom@tromey.com>, gdb-patches@sourceware.org
Subject: Re: [PATCH v2 15/22] Make exceptions use std::string and be self-managing
Date: Wed, 03 Apr 2019 16:20:00 -0000 [thread overview]
Message-ID: <f6f7893d-d3dc-6b68-adb8-88c710da19ca@redhat.com> (raw)
In-Reply-To: <20190227201849.32210-16-tom@tromey.com>
On 02/27/2019 08:18 PM, Tom Tromey wrote:
> This changes the exception's "message" member to be a std::string.
> This allows removing the stack of exception messages, because now
> exceptions will self-destruct when needed.
>
> Note that this does increase the cost of copying an exception. This
> will be somewhat addressed in a subsequent patch.
I have a concern about this patch, something that I alluded to
at the Cauldron.
I think it'd be better if we followed what the C++ standard library
does, and make copy constructor/assignment op nothrow:
"Each standard library class T that derives from class exception shall have
a publicly accessible copy constructor and a publicly accessible copy
assignment operator that do not exit with an exception." (N4659/21.8.2)
See:
https://stackoverflow.com/questions/21644735/should-i-declare-the-copy-constructor-of-my-exceptions-noexcept
I believe that rule exists because throwing an exception copies
the exception object, and if that copy operation throws, then the process
is terminated with std::terminate(). Copying a std::string may require
a heap allocation, which likely doesn't throw on most OSs with
over-commit enabled, but it pedantically can.
I think fixing this isn't hard, so I think we should do it from the
start -- instead of gdb_exception holding a std::string, hold a
reference counted immutable heap-allocated C string.
> throw_error (except.error,
> _("unable to read value of %s (%s)"),
> - xvz_name, except.message);
> + xvz_name, except.message.c_str ());
Since you're touching most (all?) lines that refer to "message",
did you consider adding a "what()" method, to model what
std::exception has?
const char *what() const noexcept;
Maybe that helps with newcomers' familiarity?
Thanks,
Pedro Alves
next prev parent reply other threads:[~2019-04-03 16:20 UTC|newest]
Thread overview: 58+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-02-27 20:18 [PATCH v2 00/22] Remove cleanups Tom Tromey
2019-02-27 20:18 ` [PATCH v2 08/22] Remove last cleanup solib-aix.c Tom Tromey
2019-03-06 19:34 ` Pedro Alves
2019-03-06 21:19 ` Tom Tromey
2019-02-27 20:18 ` [PATCH v2 05/22] Remove last cleanup from gdbserver Tom Tromey
2019-03-06 19:23 ` Pedro Alves
2019-02-27 20:19 ` [PATCH v2 13/22] Remove free_current_contents Tom Tromey
2019-03-06 21:34 ` Pedro Alves
2019-02-27 20:19 ` [PATCH v2 03/22] Change displaced_step_clear_cleanup with a forward_scope_exit Tom Tromey
2019-03-06 19:23 ` Pedro Alves
2019-02-27 20:19 ` [PATCH v2 09/22] Remove last cleanup from linux-namespaces.c Tom Tromey
2019-03-06 21:07 ` Pedro Alves
2019-02-27 20:19 ` [PATCH v2 19/22] Make exception throwing a bit more efficient Tom Tromey
2019-04-03 17:04 ` Pedro Alves
2019-02-27 20:19 ` [PATCH v2 06/22] Remove cleanup from solib-svr4.c Tom Tromey
2019-03-06 19:24 ` Pedro Alves
2019-02-27 20:19 ` [PATCH v2 04/22] C++ify remote notification code Tom Tromey
2019-03-06 19:23 ` Pedro Alves
2019-03-06 21:12 ` Tom Tromey
2019-02-27 20:19 ` [PATCH v2 14/22] Simplify exception handling Tom Tromey
2019-04-02 22:07 ` Pedro Alves
2019-02-27 20:19 ` [PATCH v2 07/22] Remove last cleanups from solib-svr4.c Tom Tromey
2019-03-06 19:32 ` Pedro Alves
2019-03-06 19:39 ` John Baldwin
2019-03-06 21:18 ` Tom Tromey
2019-02-27 20:19 ` [PATCH v2 21/22] Use SCOPE_EXIT in write_gcore_file Tom Tromey
2019-03-06 22:01 ` Pedro Alves
2019-03-06 22:31 ` Tom Tromey
2019-03-06 22:54 ` Pedro Alves
2019-03-06 22:56 ` Tom Tromey
2019-03-06 23:08 ` Tom Tromey
2019-02-27 20:19 ` [PATCH v2 20/22] Replace throw_exception with throw in some cases Tom Tromey
2019-04-03 17:05 ` Pedro Alves
2019-02-27 20:19 ` [PATCH v2 17/22] Rename gdb exception types Tom Tromey
2019-02-27 20:19 ` [PATCH v2 02/22] Update two cleanup comments Tom Tromey
2019-03-06 19:23 ` Pedro Alves
2019-02-27 20:19 ` [PATCH v2 18/22] Remove some now-dead exception code Tom Tromey
2019-04-03 17:04 ` Pedro Alves
2019-02-27 20:19 ` [PATCH v2 01/22] Remove cleanups from coffread.c Tom Tromey
2019-03-06 19:23 ` Pedro Alves
2019-02-27 20:19 ` [PATCH v2 11/22] Use unique_xmalloc_ptr in remote.c Tom Tromey
2019-03-06 21:17 ` Pedro Alves
2019-02-27 20:19 ` [PATCH v2 10/22] Remove last cleanups from stabsread.c Tom Tromey
2019-03-06 21:14 ` Pedro Alves
2019-02-27 20:19 ` [PATCH v2 15/22] Make exceptions use std::string and be self-managing Tom Tromey
2019-04-03 16:20 ` Pedro Alves [this message]
2019-04-03 17:58 ` Tom Tromey
2019-04-03 18:28 ` Pedro Alves
2019-02-27 20:19 ` [PATCH v2 16/22] Rewrite TRY/CATCH Tom Tromey
2019-04-03 17:03 ` Pedro Alves
2019-04-03 18:02 ` Tom Tromey
2019-02-27 20:19 ` [PATCH v2 12/22] Remove basic cleanup code Tom Tromey
2019-03-06 21:33 ` Pedro Alves
2019-02-27 20:19 ` [PATCH v2 22/22] Introduce and use bcache_up Tom Tromey
2019-03-06 22:03 ` Pedro Alves
2019-03-07 16:54 ` Tom Tromey
2019-03-07 17:09 ` Pedro Alves
2019-03-07 17:42 ` Tom Tromey
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=f6f7893d-d3dc-6b68-adb8-88c710da19ca@redhat.com \
--to=palves@redhat.com \
--cc=gdb-patches@sourceware.org \
--cc=tom@tromey.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox