[jira] [Comment Edited] (ROL-2101) Thread unsafe use of HashMap for cached mappings exist

["Kohei Nozaki (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/ROL-2101?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15091349#comment-15091349 ] 

Kohei Nozaki edited comment on ROL-2101 at 1/11/16 2:43 AM:
------------------------------------------------------------

I'm planning to replace the unsafe {{HashMap}} s by {{Hashtable}} as {{JPAWeblogManagerImpl}} or more recent thread safe ones ( {{ConcurrentHashMap}} or synchronized wrappers ) .

Also, I think there are some room for improvement here, And I have a question:

{code:title=JPAWeblogEntryManagerImpl#getWeblogEntryByAnchor()|borderStyle=solid}
// check cache first
// NOTE: if we ever allow changing anchors then this needs updating
if(this.entryAnchorToIdMap.containsKey(mappingKey)) {
    
    WeblogEntry entry = this.getWeblogEntry(this.entryAnchorToIdMap.get(mappingKey));
    if(entry != null) {
        LOG.debug("entryAnchorToIdMap CACHE HIT - " + mappingKey);
        return entry;
    } else {
        // mapping hit with lookup miss?  mapping must be old, remove it
        this.entryAnchorToIdMap.remove(mappingKey);
    }
}
{code}

Why this cannot be written as follows?

{code}
WeblogEntry e = this.getWeblogEntry(this.entryAnchorToIdMap.get(mappingKey));
if(e != null) {
    LOG.debug("entryAnchorToIdMap CACHE HIT - " + mappingKey);
    return e;
}
{code}

Is there posibillity of any case that {{mappingKey}} is available but value is set to null?



was (Author: xkylex):
I'm planning to replace the unsafe {{HashMap}} s by {{Hashtable}} as the above or more recent thread safe ones ( {{ConcurrentHashMap}} or synchronized wrappers ) .

Also, I think there are some room for improvement here, And I have a question:

{code:title=JPAWeblogEntryManagerImpl#getWeblogEntryByAnchor()|borderStyle=solid}
// check cache first
// NOTE: if we ever allow changing anchors then this needs updating
if(this.entryAnchorToIdMap.containsKey(mappingKey)) {
    
    WeblogEntry entry = this.getWeblogEntry(this.entryAnchorToIdMap.get(mappingKey));
    if(entry != null) {
        LOG.debug("entryAnchorToIdMap CACHE HIT - " + mappingKey);
        return entry;
    } else {
        // mapping hit with lookup miss?  mapping must be old, remove it
        this.entryAnchorToIdMap.remove(mappingKey);
    }
}
{code}

Why this cannot be written as follows?

{code}
WeblogEntry e = this.getWeblogEntry(this.entryAnchorToIdMap.get(mappingKey));
if(e != null) {
    LOG.debug("entryAnchorToIdMap CACHE HIT - " + mappingKey);
    return e;
}
{code}

Is there posibillity of any case that {{mappingKey}} is available but value is set to null?


> Thread unsafe use of HashMap for cached mappings exist
> ------------------------------------------------------
>
>                 Key: ROL-2101
>                 URL: https://issues.apache.org/jira/browse/ROL-2101
>             Project: Apache Roller
>          Issue Type: Bug
>          Components: Data Model & JPA Backend
>    Affects Versions: 5.1.2
>            Reporter: Kohei Nozaki
>            Assignee: Roller Unassigned
>            Priority: Minor
>
> There are two singleton classes that have a instance field type of {{HashMap}} for cache.
> In {{JPAUserManagerImpl}}:
> {noformat}
> // cached mapping of userNames -> userIds
> private Map<String, String> userNameToIdMap = new HashMap<String, String>();
> {noformat}
> In {{JPAWeblogEntryManagerImpl}}:
> {noformat}
> // cached mapping of entryAnchors -> entryIds
> private Map<String, String> entryAnchorToIdMap = new HashMap<String, String>();
> {noformat}
> {{HashMap}} s should not be used as the above because it's not thread safe.



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