wfsdispatch -- not cleaning up after itself.
Description
Environment
Activity
codehaus April 10, 2015 at 4:41 PM
CodeHaus Comment From: aaime - Time: Thu, 29 Mar 2007 02:02:43 -0500
---------------------
<p>This issue has been resolved for one month or more without complaints. Closing it.</p>
codehaus April 10, 2015 at 4:41 PM
CodeHaus Comment From: cholmes - Time: Mon, 13 Nov 2006 10:42:00 -0600
---------------------
<p>Ah, cool. Wish I'd found that pushback stream before. I'm pretty sure I'm to blame for that nasty hack.</p>
codehaus April 10, 2015 at 4:41 PM
CodeHaus Comment From: aaime - Time: Mon, 13 Nov 2006 10:38:07 -0600
---------------------
<p>In fact it still does, even in ows4 branch, see issue: <a href="https://jira.codehaus.org/browse/GEOS-780" title="Dispatcher should try hard not to use temp files at all..." class="issue-link" data-issue-key="GEOS-780"><del>GEOS-780</del></a>
Justin is investigating the usage of PushbackInputStream in order to avoid creation of temporary files.</p>
codehaus April 10, 2015 at 4:41 PM
CodeHaus Comment From: cholmes - Time: Mon, 13 Nov 2006 10:32:09 -0600
---------------------
<p>I believe that Justin's new dispatch system doesn't need to do this. </p>
codehaus April 10, 2015 at 4:41 PM
CodeHaus Comment From: aaime - Time: Mon, 6 Nov 2006 08:13:22 -0600
---------------------
<p>I've fixed it on 1.4.x without noticing there was an open bug. Now wfsdispatch files gets deleted as soon as the request ends.
Yet, they are still created, whilst in fact it would be better to have them not created at all... a separated issue is opened against 1.6.0 to avoid their creation.</p>
There's 1000 of "wfsdispatch36125351tmp" in my windows tmp directory. Inside each one of these is the XML for a WFS request.
This is because the WFSDispatch class isnt cleaning up after itself.
Also, consider writing the XML out to disk ONLY IF its big. Ie. write it to an in-memory temporary buffer and when that gets full write it out to disk (very similiar to the PARTIAL output strategy). Creating a a file, writing to it, reading from it, and deleting it are expensive operations. Most WFS queries are <10k, and we can stream larger ones if it necessary.