[
https://issues.apache.org/jira/browse/HADOOP-12194?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14616037#comment-14616037
]
Andrew Wang commented on HADOOP-12194:
--------------------------------------
Hi Ravi, I thought about that, but there's no easy way to enumerate the output
from the protoc generator. It ends up generating more than one class file per
input protoc, and it's not possible to determine the output file name just from
the input file name.
I think CRC is also a much safer way of detecting changes, so prefer it for
that reason as well. We already compute MD5s of all input files for the
VersionInfo plugin too (including the proto files), so doing CRC for just the
proto files shouldn't be a big overhead.
I guess one potential enhancement is sharing the checksums between the
VersionInfo and Protoc plugins, but that seems reasonable to leave for future
work.
> Support for incremental generation in the protoc plugin
> -------------------------------------------------------
>
> Key: HADOOP-12194
> URL: https://issues.apache.org/jira/browse/HADOOP-12194
> Project: Hadoop Common
> Issue Type: Improvement
> Affects Versions: 2.0.6-alpha
> Reporter: Andrew Wang
> Assignee: Andrew Wang
> Attachments: hadoop-12194.001.patch
>
>
> The protoc maven plugin currently generates new Java classes every time,
> which means Maven always picks up changed files in the build. It would be
> better if the protoc plugin only generated new Java classes when the source
> protoc files change.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
