tools/gn: provide better error message for bogus binary expressions

tools/gn: provide better error message for bogus binary expressions

BUG=472038
Committed: https://crrev.com/a2c7d0df0f3630fe4a91370df2273ebc6b4c8c49
Cr-Commit-Position: refs/heads/master@{#324106}

[mdempsky, 2015-04-06 23:38:11]

[email protected] changed reviewers:
+ [email protected]

[mdempsky, 2015-04-06 23:41:52]

https://codereview.chromium.org/1068623002/diff/1/tools/gn/parser_unittest.cc
File tools/gn/parser_unittest.cc (right):

https://codereview.chromium.org/1068623002/diff/1/tools/gn/parser_unittest.cc...
tools/gn/parser_unittest.cc:207: DoExpressionErrorTest("[a, 2+,]", 1, 7);
This expectation change is because GN used to give an error like:

    ERROR at //BUILD.gn:6:6: Expected right hand side for '+'
    [a, 2+,]
         ^

but now it instead says:

    ERROR at //BUILD.gn:6:7: Unexpected token ','
    [a, 2+,]
          ^

As mentioned in the bug, this is arguably a worse error message, but it seems to
be in-line with what C++ developers should be accustomed to from compiler error
messages for invalid input like this.

[Dirk Pranke, 2015-04-07 19:20:38]

[email protected] changed reviewers:
+ [email protected]

[Dirk Pranke, 2015-04-07 19:20:38]

This seems reasonable to me, but I'd like Scott to sign off.

[scottmg, 2015-04-07 19:49:52]

https://codereview.chromium.org/1068623002/diff/1/tools/gn/parser.cc
File tools/gn/parser.cc (right):

https://codereview.chromium.org/1068623002/diff/1/tools/gn/parser.cc#newcode391
tools/gn/parser.cc:391: if (!has_error())
nit; braces for multiline if.

https://codereview.chromium.org/1068623002/diff/1/tools/gn/parser.cc#newcode393
tools/gn/parser.cc:393: token.value().as_string() + "'");
is there a test that hits this code path now? if yes, or with that lgtm.

[mdempsky, 2015-04-07 19:58:44]

https://codereview.chromium.org/1068623002/diff/1/tools/gn/parser.cc
File tools/gn/parser.cc (right):

https://codereview.chromium.org/1068623002/diff/1/tools/gn/parser.cc#newcode391
tools/gn/parser.cc:391: if (!has_error())
On 2015/04/07 19:49:52, scottmg wrote:
> nit; braces for multiline if.

Done.

https://codereview.chromium.org/1068623002/diff/1/tools/gn/parser.cc#newcode393
tools/gn/parser.cc:393: token.value().as_string() + "'");
On 2015/04/07 19:49:52, scottmg wrote:
> is there a test that hits this code path now? if yes, or with that lgtm.

Yes, DoExpressionErrorTest("(a +", 1, 4) in Parser.ParenExpression hits this
case.  (ParseExpression(int) returns nullptr without setting an error when
called at EOF.)

[mdempsky, 2015-04-07 19:59:07]

The CQ bit was checked by [email protected]

[mdempsky, 2015-04-07 19:59:07]

The patchset sent to the CQ was uploaded after l-g-t-m from [email protected]
Link to the patchset: https://codereview.chromium.org/1068623002/#ps20001
(title: "scottmg feedback")

[commit-bot: I haz the power, 2015-04-07 20:00:09]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1068623002/20001

[commit-bot: I haz the power, 2015-04-07 20:32:07]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-04-07 20:33:53]

Patchset 2 (id:??) landed as
https://crrev.com/a2c7d0df0f3630fe4a91370df2273ebc6b4c8c49
Cr-Commit-Position: refs/heads/master@{#324106}