[jira] [Comment Edited] (YETUS-187) maven javac/javadoc can't use calcdiffs

["Allen Wittenauer (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/YETUS-187?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15060105#comment-15060105 ] 

Allen Wittenauer edited comment on YETUS-187 at 12/16/15 3:09 PM:
------------------------------------------------------------------

Neat. Found a bug in shellcheck. :/

Related: https://github.com/koalaman/shellcheck/issues/518


was (Author: aw):
Neat. Found a bug in shellcheck. :/

> maven javac/javadoc can't use calcdiffs
> ---------------------------------------
>
>                 Key: YETUS-187
>                 URL: https://issues.apache.org/jira/browse/YETUS-187
>             Project: Yetus
>          Issue Type: Bug
>          Components: Test Patch
>    Affects Versions: 0.1.0
>            Reporter: Allen Wittenauer
>            Assignee: Allen Wittenauer
>         Attachments: YETUS-187.00.patch
>
>
> calcdiffs works by doing the following:
> * given two files (A, B)
> * from (A,B), pull everything from the last colon(\:) out and into (C,D). This gives us a list of all errors in order without line numbers, columns, or filenames in the way.
> * do a diff of C and D, printing the numbers of the lines that are different from D
> * using the line numbers, pull out the lines from B, restoring the filenames, lines, and columns
> End result will be, the vast majority of the time (but not 100%), the new errors generated by B. The places where this doesn't work is when the surrounding context isn't different enough. e.g., multiple messages with the same text.  There obviously could be some improvement in accuracy here, but experience has shown this works well enough the majority of the time.
> For maven (and maybe other tools?), though, the place where this falls down is the line format:
> {code}
> [WARNING] Endpoint.java:[162,25] [unchecked] Possible heap pollution from parameterized vararg type Map<String,String>
> {code} 
> In this situation, we end up doing the diff with the line numbers and columns in place which means we're very likely to print any pre-existing errors too.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)