This reverts commit 65e294e221e0ad1d37c6667eecb995817ace870d, which
breaks several widely used Basic extensions and hundreds of anonymous
macros that miss closing parenthesis here and there
(cherry picked from commit 660e394b893a045a077ccded263f94a1be03fcf9)
Change-Id: I5d27e6d4e7c8aea3f0762c41ede46e35a5a24aaa
Reviewed-on: https://gerrit.libreoffice.org/35970
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>
regression from
commit f17a4694b07856292804c23b80ce92967d401bb8
Author: Arnaud Versini <arnaud.versini@gmail.com>
Date: Sat Feb 13 20:17:41 2016 +0100
BASIC : use std::unique_ptr for storing SbiExprList
Change-Id: I37f2a1c837c6742cf6d403962d2730b5e80004ec
Reviewed-on: https://gerrit.libreoffice.org/22345
because they have been moved into the container at this
point
Change-Id: If21e0321ffdce66aa9629637d0ff8c16efbbd25e
...4b4a7c0d87eb580272aba0777c9021789025bdc0 "Revert a fix that can never have
worked in the first place?"---which presumably did work after all, as Noel
thankfully pointed out to me,
<http://lists.freedesktop.org/archives/libreoffice/2015-September/070193.html>
"Re: [Libreoffice-commits] core.git: basic/source 'Revert a fix that can never
have worked in the first place?'"
Change-Id: I411bfaacbfebf50589290c6a3040d0300c256439
clang-analyzer-deadcode.DeadStores finds that conditionally assigning
eTok = SYMBOL
inside the "if( SbiTokenizer::IsKws( eTok ) )" block is useless, as the directly
following
if( DoParametersFollow( pParser, eCurExpr, eTok = eNextTok ) )
will unconditionally assign into eTok again, without intermediate use of the old
eTok value.
Now, the conditional "eTok = SYMBOL" assignment was added as
5d98ed5c6a4afc0a7943318c510e56aef8c45193 "INTEGRATION: CWS ab12fixes: #118234#
SbiExpression::Term(): Allow Input as symbol for action option compatible,"
while the unconditional "eTok = eNextTok" assignment had followed that block
ever since c25ec0608a167bcf1d891043f02273761c351701 "initial import."
The referenced "#118234#" was a Sun-internal bug and is no longer available, but
it does very much look as if this alleged bugfix never worked in the first
place. So revert the code back to what it looked before
5d98ed5c6a4afc0a7943318c510e56aef8c45193, for now.
Change-Id: I1fe1178d2c5b0c0372da32b8dd0f2dfbdb22a1ae
we can only enter the bAssumeExprLParenMode controlled block once
because the block sets bAssumeExprLParenMode to false
bAssumeExprLParenMode is only true if the first token read in the
method is LPAREN
so we can only enter this block if the current token is LPAREN and
its the first token processed, so the token can't be BYVAL
Change-Id: I6e668fca4d127d7dbfe447e2d9f2231e05278a7d
parsing 1000s of line of code is hard enough without having to fight
with weird indentation and irregular formatting.
So as the review progress, in order to follow the code, cosmetic changes
were made...
In order to minimize the task of the reviewers and allow them to
concentrate on what matter, an effort is made to collect these
cosmetic changes into this separate commit.
Change-Id: I3c9b04a0150d0d0a048c2e976fe24de4f2b6b98a
