DEBUG [SequenceGenerator] Sequence identifier generated: 1025
DEBUG [AbstractBatcher] about to close PreparedStatement (open PreparedStatements: 1, globally: 1)
DEBUG [ConnectionManager] aggressively releasing JDBC connection
DEBUG [ConnectionManager] releasing JDBC connection [ (open PreparedStatements: 0, globally: 0) (open ResultSets: 0, globally: 0)]
DEBUG [AbstractSaveEventListener] generated identifier: 1025, using strategy: org.hibernate.id.SequenceGenerator
DEBUG [AbstractSaveEventListener] generated identifier: 1025, using strategy: org.hibernate.id.ForeignGenerator
DEBUG [AbstractFlushingEventListener] processing flush-time cascades
DEBUG [AbstractEntityManagerImpl] mark transaction for rollback

It’s a one-to-one foreign key generated by a DB2 sequence to persist multiple entities in one transaction. It seems to be created and assigned to the two entities mapped to each other but I have no clue why the JPA/Hibernate transaction decides it’s a no go as all the required fields are assigned in the objects being persisted. I’ve been working on this Seam Framework wizard for a week and this is an example of roadblocks in the dozens that have taken me to varying edges of sanity when it comes to troubleshooting this project.

I can only take solace in knowing once these niggles are straightened out, the rest of the system will be easy to implement. But the technological learning curve and unseen issues derived from requirement complexities make me believe Seam is a terrible framework solution. Absolutely terrible.

Update July 2: This error was occurring because a RunTimeException was thrown in the Hibernate library which wasn’t displaying in the server’s log or on the JBoss stack trace because the entrance method was invoked using EL. Wrapping my persist() call with a try {} containing a generic Exception catch revealed it was a constraint violation occurring on a value set as a Hibernate TrueFalseType (char(1) with ‘T’ or ‘F’ values) when it should’ve been a YesNoType (char(1) with ‘Y’ or ‘N’ values).

You can’t. But you can use <h:outputText> for anonymous visitors (read-only) and <rich:inplaceInput> for known logged-in users (write-enabled.) Customize the inplaceInputTest.xhtml s:hasRole() parameter to give the proper boolean value on whether the logged-in user is an administrator or not of the field bound to the input.

Define a custom tag to replace <rich:inplaceInput>

/resources/WEB-INF/web.xml

.
.
.
  
    facelets.LIBRARIES
    /WEB-INF/compositions.taglib.xml
  

/resources/WEB-INF/compositions.taglib.xml

  http://mytaglib.com/jsf
  
    inplaceInput
    ../compositions/inplaceInput.xhtml
  

Below, evaluation of the rendered="" attribute decides which of the two elements appear at one time.

/view/compositions/inplaceInput.xhtml

Custom <mytaglib:inplaceInput> Usage

/view/inplaceInputTest.xhtml

Yeah, it’s all happening. Migrating a CMS to Java’s Seam framework using an existing DB2 schema/dataset, the Lucene search indexes on a few tables caused DB2 error SQL1584N. The bizarre part was it was only happening locally on my machine, but not for two others at work, who each have the same amount of RAM. It may have been because I was on Ubuntu and the other devs use Red Hat Enterprise Linux.

System temporary tablespace with page size of at least pagesize could not be found.
Explanation:
A system temporary table space was required to process the statement. There was no system temporary table space available that had a page size of pagesize or larger.

The statement cannot be processed.

User Response:
Create a system temporary table space with a page size of at least pagesize.

sqlcode: -1584

sqlstate: 57055

Following the instructions for ensuring system temporary table spaces page sizes meet requirements, I ran DB2 query:

LIST TABLESPACES SHOW DETAIL

…checking for the entry with Type = System managed space, Contents = System Temporary data, I confirmed the page size used was 32k, which is the highest possible with DB2.

After googling a bit, it seemed the solution was to increase the bufferpool size used by the tablespace giving me this error. The instructions for adjusting the bufferpool are to simply run DB2 query:

SELECT * FROM syscat.bufferpools

Confirm the bufferpool size, then increase the number of pages available until the Lucene indexing doesn’t shit the bed. e.g.,

ALTER BUFFERPOOL ibmdefaultbp SIZE 2000

During this process, you may get the error:

SQL20169W The buffer pool is not started.

sqlcode: +20169

sqlstate: 01654

This just means the change won’t take effect until DB2 is restarted, so execute:

db2 force applications all
db2stop
db2start

Now, if you have the Seam observer Java annotation setup for Lucene indexing, this DB2 error will begone on the next deploy. And yes, this whole post makes me yearn for a non-Java project started from scratch.

The Game Informer reveal of Portal 2 details led to speculation that new gameplay elements will follow the same path as Digipen student project Narbacular Drop (which the original Portal was inspired by). Colour me as again missing the boat, but the 2008 release of Tag (~50mb download) passed me by. It was a first-person puzzle platformer with a goal to simply navigate through levels. It’s really a tech demo, but its possibilities really highlight how to improve Portal‘s mechanics… and it just so happens its creators may now be under Valve’s employ.

The fun core of the game is manipulating surfaces through use of a paintball gun to allow jumping, speeding up your running speed, and allowing you to pull off a gravity-defying walks on any surface (like Prey‘s ceiling-traversal). The open-air urban setting (used due to the jumping component) recalls Mirror’s Edge environments but with monochrome, texture-free geometry combined with a graffiti aesthetic. Check it out if you want to see what Portal 2 may play like at year’s end.

There are many robust JavaScript libraries out there for handling timestamps, but I had a situation requiring a lightweight solution to display dates returned from a JSON request. I wanted human-readable, contextual time to indicate the proximity to an expiration. Something to the effect of, “Feb 3, 2010 (in 5 days)”, or, “Jan 1, 2010 (30 days ago)”.

This implementation doesn’t handle language localization or customizing messages like optionally breaking down to also include years or decades. But it does contain nested ugliness with magic numbers ahoy! I didn’t say it was elegant; just lightweight (see: took less than 10 minutes to create, even with clever pluralization of each unit.)

/**
 * Get a string describing the difference between two timestamps.
 *
 * @var {Date} time To retrieve relative descriptive string for.
 *
 * @return string indicating how far in the past or future the date is
 */
function getTimeDiff(time)
{
    var diff = (new Date()).getTime() - time.getTime();

    return getTimeDiffDescription(diff, 'day', 86400000) ||
        getTimeDiffDescription(diff, 'hour', 3600000) ||
        getTimeDiffDescription(diff, 'minute', 60000) ||
        getTimeDiffDescription(diff, 'second', 1000) ||
        'just now';
}

/**
 * Get a string describing the difference between two timestamps,
 * based on one unit. Run sequentially starting from the greatest
 * unit and stopping once this function doesn't return null.
 *
 * @var diff current time (in millisec since 1970) minus time to compare to.
 *           e.g., (new DateTime()).getTime() - someOtherTime.getTime()
 * @var unit Time description to place in a string. e.g., 'hour'
 * @var divisor Number to divide into milliseconds to get the value
 *              described by the unit. e.g., 60*60*1000 for 'hour'
 *
 * @return A string representing the difference between the given time
 *         and now, if there is a difference for the given unit of time.
 *         null is returned if they're the same (e.g., provided time falls
 *         on today when unit = 'day')
 */
function getTimeDiffDescription(diff, unit, divisor)
{
    var unitAmount = (diff / divisor).toFixed(0);

    if (unitAmount > 0) {
        return unitAmount + ' ' + unit + (unitAmount == 1 ? '': 's') + ' ago';
    } else if (unitAmount < 0) {
        return 'in ' + Math.abs(unitAmount) + ' ' + unit + (unitAmount == -1 ? '' : 's');
    } else {
        return null;
    }
}

Edit from 2016: In all fairness, Moment.js is your best bet.

After owning a Sony Ericsson K850i for 18 months (and under contract until June 2012), I thought it was time to get my contacts in order. Finding a seamless method to keep the phone, iPod Touch, and Hotmail contacts all synchronized wasn’t easy. My main stumbling block was me being a stickler for a non-Gmail email address I’ve been using since 1997, but a solution must exist.

I had to install the Sony Ericsson PC Suite which promptly gave the message:

Reading mobile phone’s contacts failed.

Few relevant Google results match my problem other than this post on sepcsuite.exe. It claims you need to install drivers (which the PC Suite does.) The error message is nowhere in Ericsson’s support pages.

A year ago, I disabled phone modem-related drivers because every USB recharge caused five “new hardware installed” notifications. Every time. So I re-enabled those in Device Manager and also reinstalled the PC Suite before my phone becomes recognized for syncing.

Google Sync supports over-the-air Gmail contacts but I don’t have a data plan so I’m not touching that given per-kilobyte pricing in Canada. Syncing by USB, Thunderbird isn’t an option; only MS Outlook. So my solution was to download a Hotmail .csv, import into MS Outlook Express (like I’d pay for MS Office), then use the PC Suite for the final sync.

Problem is I still had contacts on my SIM the software can’t remove. The Sony Ericsson K850i has no method on the phone itself to batch delete contacts from a SIM. I had to do that one-by-one or else there were redundant contacts for most people. Sony Ericsson’s OS has latency issues, so it was unbelievably laggy on 165 contacts. I’d wait half a second between deletes for the contacts list to (twice) refresh. Then PC Suite offered no option to only sync contacts with (mobile, work, or home) phone numbers. I had to manually remove entries under the software’s Contacts tab which has a phone number table in view. So in the end, I found no elegant solution to sync contacts.

Some thoughts on last night’s The Twilight Sad show in Toronto:

• Venue was a dudefest. There was a line-up for the men’s urinals and an empty ladies washroom. I have only seen this at large-scale sporting events.
• The bartenders kept giving me a handful of change expecting me to tip them higher. When I give you a $10 bill for a $5.25 bottled beer, don’t hand me 3 quarters and 4 loonies. That means I will tip you even less. $0.75 for you.
• I was slightly right of centre six people back and could barely hear the lead guitarist Andy MacFarlane positioned to the left. Of course, it was because the amps were stacked behind him for controlled feedback but there should’ve been something mic’d so the whole crowd could get a better mix. His wailing is their signature sound!
• They played for about an hour with no encore. Who are you? Tool? I don’t believe their set was any longer than when they opened for Mogwai back in May.
• Their bassist may be a robot.
• Love the live interpretation of Fourteen Autumns and Fifteen Winters’ first song “Cold Days From the Birdhouse” instead having an a capella intro.
• Fellow Scots We Were Promised Jetpacks are pretty awesome. Glad I accidentally arrived early enough to catch their set.

All points aside, solid show with much love around.

PayPal has a sandbox development environment to test project requirement flows, but instructions found in the documentation and forums are consistently vague and at times categorically incorrect. If troubleshooting documentation leads to cycles around the same pages, it’s a problem. For instance, this message I received when attempting to process a GetRecurringPaymentsProfileDetails request:

PayPal error code: 11546 “Profile ID is not valid for this account. Please resubmit request with the correct profile ID.”

As Conan’s time winds down before moving in Leno’s slot, I think I’ll most miss the obscene bit characters and hilariously inappropriate on-site shoots. Last night’s episode contained yet another terrific sequence involving alcohol and an awkward Irish man, but the episode went into side-splitting territory once Norm MacDonald and Gordon Ramsay got together. Late Night always had a low-budget vibe to it that allowed anything goes, so it’ll be disheartening if his humous has to be toned down to fit in with the pedestrian Leno demographic (hey, tell another fecking Lewinsky joke!)