[PATCH 3/5] Add (more) support for WIN32 attribute names

[PATCH 3/5] Add (more) support for WIN32 attribute names

[Ramsay Jones]

In particular, the following identifiers (along with their __X__ variants)
are now accepted as attribute names: fastcall, dllimport and dllexport.
(cdecl and stdcall were added in baf2c5a84e by Michael Stefaniuc).
For now, at least, these attributes are just ignored.

Signed-off-by: Ramsay Jones <ramsay@ramsay1.demon.co.uk>
---
  ident-list.h |    7 +++++++
  parse.c      |    6 ++++++
  2 files changed, 13 insertions(+), 0 deletions(-)

diff --git a/ident-list.h b/ident-list.h
index 7633a2f..d09c2ab 100644
--- a/ident-list.h
+++ b/ident-list.h
@@ -70,6 +70,13 @@ IDENT(nonnull); IDENT(__nonnull); IDENT(__nonnull__);
  IDENT(__deprecated__);
  IDENT(__warn_unused_result__);

+/* WIN32 specific Attribute names */
+IDENT(cdecl); IDENT(__cdecl__);
+IDENT(stdcall); IDENT(__stdcall__);
+IDENT(fastcall); IDENT(__fastcall__);
+IDENT(dllimport); IDENT(__dllimport__);
+IDENT(dllexport); IDENT(__dllexport__);
+
  /* Preprocessor idents */
  __IDENT(pragma_ident, "__pragma__", 0);
  __IDENT(__VA_ARGS___ident, "__VA_ARGS__", 0);
diff --git a/parse.c b/parse.c
index b74c7c0..db5c9e6 100644
--- a/parse.c
+++ b/parse.c
@@ -327,6 +327,12 @@ static struct init_keyword {
  	{ "__cdecl__",	NS_KEYWORD,	.op = &ignore_attr_op },
  	{ "stdcall",	NS_KEYWORD,	.op = &ignore_attr_op },
  	{ "__stdcall__",	NS_KEYWORD,	.op = &ignore_attr_op },
+	{ "fastcall",	NS_KEYWORD,	.op = &ignore_attr_op },
+	{ "__fastcall__",	NS_KEYWORD,	.op = &ignore_attr_op },
+	{ "dllimport",	NS_KEYWORD,	.op = &ignore_attr_op },
+	{ "__dllimport__",	NS_KEYWORD,	.op = &ignore_attr_op },
+	{ "dllexport",	NS_KEYWORD,	.op = &ignore_attr_op },
+	{ "__dllexport__",	NS_KEYWORD,	.op = &ignore_attr_op },
  };

  void init_parser(int stream)
-- 
1.5.1

Re: [PATCH 3/5] Add (more) support for WIN32 attribute names

[Josh Triplett]

[-- Attachment #1: Type: text/plain, Size: 961 bytes --]

Ramsay Jones wrote:
> In particular, the following identifiers (along with their __X__ variants)
> are now accepted as attribute names: fastcall, dllimport and dllexport.
> (cdecl and stdcall were added in baf2c5a84e by Michael Stefaniuc).
> For now, at least, these attributes are just ignored.

Your patch seems to have some whitespace damage in the form of extra leading
spaces.  I've manually modified it to apply this time, but please figure out
what causes these patches to break and fix it.  You may want to test by
sending patches to yourself and trying to apply them.  I notice that you use
Thunderbird; format=flowed may cause the problem, so try turning it off by
disabling the preference mailnews.send_plaintext_flowed .  You can do so
either via Edit -> Preferences -> Advanced -> Config Editor; alternatively, if
you have Enigmail installed, it offers a checkbox to turn off format=flowed in
its preferneces.

- Josh Triplett



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 252 bytes --]

Re: [PATCH 3/5] Add (more) support for WIN32 attribute names

[Ramsay Jones]

Josh Triplett wrote:
> Your patch seems to have some whitespace damage in the form of extra leading
> spaces.  I've manually modified it to apply this time, but please figure out
> what causes these patches to break and fix it.  You may want to test by
> sending patches to yourself and trying to apply them.  I notice that you use
> Thunderbird; format=flowed may cause the problem, so try turning it off by
> disabling the preference mailnews.send_plaintext_flowed .  You can do so
> either via Edit -> Preferences -> Advanced -> Config Editor; alternatively, if
> you have Enigmail installed, it offers a checkbox to turn off format=flowed in
> its preferneces.
> 

Indeed, all 5 patches have whitespace damage ;-(
Sorry for messing that up. I was under the (false) impression that I had knocked
Thunderbird into submission on this issue; my git patch submissions had not been
rejected, so I just assumed everything was OK. However, it appears that, after
inspecting the git patch e-mails left in my sent folder, all my git patches have
suffered the same problem. (ie Junio has been silently fixing them up! - oops).

After much study last night, it seems that the pattern of corruption is: for all
lines that start with a space, insert an extra space, except for lines which
consist of a single space, which is removed instead. 8-)

However, it seems that after setting "mailnews.send_plaintext_flowed" as you
suggest above, the extra space is no longer being inserted. The elimination of
the lone space is still happening, but git-apply seems to be OK with it!

I have regenerated the patch e-mails, done a "Save As" from the unsent folder,
converted the resulting *.eml files to unix line ending and verified that they
apply using:

    $ git apply --check --verbose --stat $patch-file

NOTE: git-apply will fail if the *.eml files have cr-lf line ending.

I hope you don't mind if I resend the e-mails to confirm they have been fixed.
(I have removed the mailing-list from the cc:, so as not to spam the list)

By the way, the second patch did not reach the list, because it had three
uppercase x chars in the subject and was bounced by the vger spam blocker ;-)

ATB,

Ramsay Jones

Re: [PATCH 3/5] Add (more) support for WIN32 attribute names

[Randy Dunlap]

On Thu, 24 May 2007 16:59:03 +0100 Ramsay Jones wrote:

> Indeed, all 5 patches have whitespace damage ;-(
> Sorry for messing that up. I was under the (false) impression that I had knocked
> Thunderbird into submission on this issue; my git patch submissions had not been
> rejected, so I just assumed everything was OK. However, it appears that, after
> inspecting the git patch e-mails left in my sent folder, all my git patches have
> suffered the same problem. (ie Junio has been silently fixing them up! - oops).
> 
> After much study last night, it seems that the pattern of corruption is: for all
> lines that start with a space, insert an extra space, except for lines which
> consist of a single space, which is removed instead. 8-)
> 
> However, it seems that after setting "mailnews.send_plaintext_flowed" as you
> suggest above, the extra space is no longer being inserted. The elimination of
> the lone space is still happening, but git-apply seems to be OK with it!

This may help with tbird:
http://mbligh.org/linuxdocs/Email/Clients/Thunderbird

---
~Randy
*** Remember to use Documentation/SubmitChecklist when testing your code ***