[
https://issues.apache.org/jira/browse/FILEUPLOAD-168?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12639439#action_12639439
]
Simon Kitching commented on FILEUPLOAD-168:
-------------------------------------------
It seems to me that *something* needs to be done at least.
As the size is being checked even in streaming mode there is no reasonable way
to access data after the file body even in streaming mode. Setting sizeMax to
an artificially-high level then implementing a custom size check while reading
from the stream is possible as a workaround, but is really pretty ugly.
I think the current behaviour should remain the default - AFACT most people
will *not* need the params that follow the file to be available (or there would
have been complaints already). Obviously reading all the file contents from the
stream (and just discarding them) is always inefficient, and for most people
not needed. But clearly for some people it is necessary.
I'm not sure about the attached patch. Simply continuing to return data when
the size is exceeded seems pointless as that data is clearly not needed.
Instead, how about detecting size overflow then simply entering a loop to
read-and-discard-input until end of that mime part is found, then throw the
exception. People who care could then catch the exception and keep going as the
input stream is correctly positioned.
> read form field parameters even if maxSize has been exceeded
> ------------------------------------------------------------
>
> Key: FILEUPLOAD-168
> URL: https://issues.apache.org/jira/browse/FILEUPLOAD-168
> Project: Commons FileUpload
> Issue Type: Improvement
> Environment: commons fileupload 1.2.2-SNAPSHOT
> Reporter: Paul Rivera
> Attachments: fileupload1.patch, fileupload2.patch
>
>
> Hi!
> This issue is similar to FILEUPLOAD-140. I can't seem to reopen it so I
> created a new one instead. FILEUPLOAD-140 was marked as resolve by using the
> streaming API, if I'm not mistaken. No change was done. But I disagree on
> the resolution of simply using the streaming API (as detailed in
> http://commons.apache.org/fileupload/streaming.html).
> First of all, I tried to upload a big file exceeding maxSize with streaming
> API and got the SizeLimitExceededException even before ANY parameter has been
> read.
> ServletFileUpload.getItemIterator() calls FileUploadBase.getItemIterator()
> which creates a new FileItemIteratorImpl(). In the constructor code of
> FileItemIteratorImpl, it already checks for the requestSize and throws
> SizeLimitExceededException if sizeMax is exceeded.
> I'd like to open this discussion again and hope that in the end, we can have
> either:
> - form field parameters BEFORE the file parameter will still get read if
> requestSize is greater than sizeMax and then terminate once we reach the file
> - all form field parameters will still get read if requestSize is greater
> than sizeMax. But, we should skip reading the body of the files and proceed
> to the next 'boundary' so as not to keep the user waiting, if ever this is
> possible. (preferred)
> Then, we should also apply the same improvement into PortletFileUpload.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
