[
https://issues.apache.org/jira/browse/THRIFT-3481?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15054932#comment-15054932
]
Aki Sukegawa commented on THRIFT-3481:
--------------------------------------
As you already noted, the functionality is very similar to existing one.
Did you consider adding a new policy to TMemoryBuffer ?
Honestly IMHO, OBSERVE should have been WRAP or something that prevents realloc
but not writing.
It tries to deal with memory ownership and constness at once, turning latter
into run-time error rather than compile-time.
The error message "Insufficient space" is misleading too for that matter...
That said, adding another policy without breaking any backward compatibility
seems feasible with a few modifications.
What do you think ?
> introduce TSimpleMemoryTransport which wraps a customer owned memory chunk to
> conduct both read and write operations
> --------------------------------------------------------------------------------------------------------------------
>
> Key: THRIFT-3481
> URL: https://issues.apache.org/jira/browse/THRIFT-3481
> Project: Thrift
> Issue Type: New Feature
> Components: C++ - Library
> Affects Versions: 0.9.3
> Reporter: Xiaoshuang LU
> Fix For: 0.9.4
>
> Attachments: THRIFT-3481.patch
>
>
> It would be great if there is a transport class which 1. does not need to
> care about the memory management, 2. is capable of both read and write
> operations.
> TMemoryBuffer is very close to the requirements, but no data can be written
> to observed TMemoryBuffer as stated in the comments.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
