From: Tom Tromey <tromey@redhat.com>
To: Trevor Saunders <tsaunders@mozilla.com>
Cc: gcc-patches@gcc.gnu.org
Subject: Re: [PATCH 5/5] add libcc1
Date: Fri, 20 Jun 2014 15:34:00 -0000 [thread overview]
Message-ID: <87vbrv8uje.fsf@fleche.redhat.com> (raw)
In-Reply-To: <20140620031058.GA12633@tsaunders-iceball.corp.tor1.mozilla.com> (Trevor Saunders's message of "Thu, 19 Jun 2014 23:10:58 -0400")
>>>>> "Trevor" == Trevor Saunders <tsaunders@mozilla.com> writes:
Trevor> I'm curious, what is the reason you choose not to write this in C++11 or
Trevor> later? Its distributed with gcc, so the only case where you aren't
Trevor> building with the in tree compiler and libraries is when your cross
Trevor> compiling gcc, and it doesn't seem particularly important to support
Trevor> building the plugin or library in that configuration. So istm you could
Trevor> go all the way and assume you are being built with trunk gcc and
Trevor> libraries.
The plugin has to be ABI compatible with GCC itself, and my
understanding was that C++11 and "however GCC is built" are not
necessarily compatible.
Switching to C++11 would be an improvement -- variadic templates would
simplify the RPC code (with a complicated caveat). So if it is possible
I am interested.
Trevor> I'm going to use this as an excuse to bring up something I've wanted to
Trevor> discuss for a while.
Trevor> So can we add C++ stuff to libiberty and allow building
Trevor> libiberty without it for binutils / gdb, or can we do something
Trevor> else to avoid this kind of stuff?
One way would be to just make a new top-level directory for a new
library.
Trevor> This question also arises in the case of templating splay_tree, and I
Trevor> imagine if gdb switches to C++ some day they'll want to reuse vec.h.
While I would like that to happen, I think the odds are very long now.
>> + connection (int fd)
>> + : m_fd (fd),
>> + m_aux_fd (-1),
>> + m_callbacks ()
Trevor> Personally I'd leave that to the compiler to write, but I guess there's
Trevor> something to be said for being explicit.
I can't recall if I did this in response to a warning or if it was just
because I wanted to be explicit.
I'm inclined to leave it, but I suppose only out of inertia.
>> + void print (const char *buf)
Trevor> explicitly mark it as virtual?
Good idea, done.
>> +// This is a wrapper function that is called by the RPC system and
>> +// that then forwards the call to the library user. Note that the
>> +// return value is not used; the type cannot be 'void' due to
>> +// limitations in our simple RPC.
>> +gcc_address
Trevor> looks like this one probably is used?
Thanks, fixed.
>> + char **argv = new (std::nothrow) char *[self->args.size () + 1];
Trevor> What's the point of making this no throw? you don't null check it so
Trevor> you'll crash anyway afaict.
Thanks. I changed it to do a NULL check. It's nothrow because nothing
in libcc1 or gdb is prepared for a C++ exception. While I like
exceptions (gdb uses its own longjmp-based exception system
extensively), my understanding is that they aren't currently used in
gcc.
>> +cc1_plugin::status
>> +cc1_plugin::unmarshall (connection *conn, char **result)
>> +{
>> + unsigned long long len;
>> +
>> + if (!conn->require ('s'))
>> + return FAIL;
>> + if (!conn->get (&len, sizeof (len)))
>> + return FAIL;
>> +
>> + if (len == -1ULL)
>> + {
>> + *result = NULL;
>> + return OK;
>> + }
>> +
>> + char *str = new (std::nothrow) char[len + 1];
Trevor> It'd be really nice if the type of the out arg forced the caller to deal
Trevor> with deleting the string like unique_ptr<char>, it would be even nicer
Trevor> if you could stick a random buffer in a std::string, but I guess you
Trevor> can't :(
Yeah, it's all quite simplistic. I suppose it could be upgraded, there
just didn't seem to be a need.
Trevor> Also where does this array get deleted?
The unmarshalling methods are generally called via argument_wrappers.
An example is in connection.cc:
// Use an argument_wrapper here to simplify management
// of the string's lifetime.
argument_wrapper<char *> method_name;
if (!method_name.unmarshall (this))
return FAIL;
Then in rpc.hh:
// Specialization for string types.
template<>
class argument_wrapper<const char *>
{
public:
argument_wrapper () : m_object (NULL) { }
~argument_wrapper ()
{
delete[] m_object;
}
Tom
next prev parent reply other threads:[~2014-06-20 15:34 UTC|newest]
Thread overview: 101+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-05-16 15:26 [PATCH 0/5] let gdb reuse gcc'c C compiler Tom Tromey
2014-05-16 15:26 ` [PATCH 1/5] export finish_bitfield_layout from stor-layout Tom Tromey
2014-05-16 18:27 ` Jeff Law
2014-05-16 15:27 ` [PATCH 4/5] add gcc/gdb interface files Tom Tromey
2014-05-22 12:52 ` Jeff Law
2014-05-22 13:16 ` Jakub Jelinek
2014-05-22 15:13 ` Jeff Law
2014-06-04 20:39 ` Tom Tromey
2014-06-05 19:23 ` Jeff Law
2014-06-05 19:30 ` Jakub Jelinek
2014-06-06 1:54 ` Tom Tromey
2014-06-06 5:35 ` Jakub Jelinek
2014-06-09 17:09 ` Jeff Law
2014-06-19 20:45 ` Tom Tromey
2014-05-16 15:27 ` [PATCH 2/5] c_diagnostic_ignored_function hack Tom Tromey
2014-05-16 16:05 ` Joseph S. Myers
2014-06-19 20:46 ` Tom Tromey
2014-05-16 18:33 ` Jeff Law
2014-05-16 18:42 ` Tom Tromey
2014-05-16 15:27 ` [PATCH 3/5] introduce the binding oracle Tom Tromey
2014-05-16 19:12 ` Jeff Law
2014-06-05 15:31 ` Tom Tromey
2014-06-19 20:43 ` Tom Tromey
2014-05-16 16:29 ` [PATCH 0/5] let gdb reuse gcc'c C compiler Mike Stump
2014-05-16 18:48 ` [PATCH 5/5] add libcc1 Tom Tromey
2014-05-16 20:29 ` Joseph S. Myers
2014-05-16 21:03 ` Tom Tromey
2014-06-18 21:07 ` Tom Tromey
2014-06-19 20:47 ` Tom Tromey
2014-05-16 21:36 ` Mike Stump
2014-05-16 22:17 ` Mike Stump
2014-06-05 19:34 ` Tom Tromey
2014-06-05 21:35 ` Mike Stump
2014-06-19 20:52 ` Tom Tromey
2014-06-19 21:45 ` Jakub Jelinek
2014-06-19 22:22 ` Tom Tromey
2014-06-20 3:11 ` Trevor Saunders
2014-06-20 15:34 ` Tom Tromey [this message]
2014-06-23 19:09 ` Jeff Law
2014-06-24 3:13 ` Trevor Saunders
2014-06-24 17:12 ` Tom Tromey
2014-06-24 18:10 ` Trevor Saunders
2014-07-18 19:00 ` Tom Tromey
2014-07-31 4:49 ` Jeff Law
2014-07-31 8:15 ` Richard Biener
2014-07-31 10:53 ` Trevor Saunders
2014-07-31 11:28 ` Richard Biener
2014-07-31 15:08 ` Joseph S. Myers
2014-07-31 19:20 ` Tom Tromey
2014-07-31 19:21 ` Jakub Jelinek
2014-07-31 19:51 ` Trevor Saunders
2014-07-31 20:07 ` Tom Tromey
2014-08-01 2:18 ` Trevor Saunders
2014-07-31 21:14 ` Mike Stump
2014-08-04 14:23 ` Tom Tromey
2014-08-05 19:34 ` Tom Tromey
2014-08-08 12:15 ` [PATCH 5/5] add libcc1 [gcc-5/changes.html] Jan Kratochvil
2014-08-31 15:12 ` Gerald Pfeifer
2014-09-14 15:07 ` Manuel López-Ibáñez
2014-09-17 14:14 ` Jan Kratochvil
2014-10-09 9:07 ` [PATCH 5/5] add libcc1 Phil Muldoon
2014-10-09 9:12 ` Jakub Jelinek
2014-10-09 9:18 ` Phil Muldoon
2014-10-10 22:31 ` Jeff Law
2014-10-24 7:43 ` Phil Muldoon
2014-10-24 7:53 ` Jakub Jelinek
2014-10-24 16:47 ` Jeff Law
2014-10-27 20:03 ` Phil Muldoon
2014-10-28 13:29 ` Joseph S. Myers
2014-10-28 18:00 ` Phil Muldoon
2014-10-29 3:32 ` Joseph S. Myers
2014-10-29 10:29 ` Jakub Jelinek
2014-10-29 10:45 ` Paolo Bonzini
2014-10-29 10:58 ` Jakub Jelinek
2014-10-29 10:59 ` Paolo Bonzini
2014-10-29 11:03 ` Jakub Jelinek
2014-10-29 15:00 ` Paolo Bonzini
2014-10-29 11:10 ` Phil Muldoon
2014-10-29 11:42 ` Phil Muldoon
2014-10-30 5:37 ` Jeff Law
2014-06-19 20:42 ` [PATCH 0/5] let gdb reuse gcc'c C compiler Tom Tromey
2014-07-30 16:28 ` Tom Tromey
2014-10-28 3:19 [PATCH 5/5] add libcc1 Dominique Dhumieres
2014-10-28 8:36 ` Jakub Jelinek
2014-10-28 8:46 ` Phil Muldoon
2014-10-28 8:50 ` Christophe Lyon
2014-10-28 9:09 ` Jakub Jelinek
2014-10-28 8:43 Uros Bizjak
2014-10-28 8:55 ` Phil Muldoon
2014-10-28 9:21 ` Uros Bizjak
2014-10-28 9:36 ` Jakub Jelinek
2014-10-28 9:58 ` Uros Bizjak
2014-10-28 10:51 ` Phil Muldoon
2014-10-28 10:56 ` Uros Bizjak
2014-10-28 11:16 ` Phil Muldoon
2014-10-28 11:53 ` Phil Muldoon
2014-10-28 12:24 ` Jakub Jelinek
2014-10-30 5:16 ` Jeff Law
2014-10-30 8:50 ` Thomas Schwinge
2014-10-30 8:56 ` Jakub Jelinek
2014-10-30 10:43 ` Thomas Schwinge
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=87vbrv8uje.fsf@fleche.redhat.com \
--to=tromey@redhat.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=tsaunders@mozilla.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;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).