URL url = new URL( myAwesomeURLString ); InputStream in = url.openStream();
In the context of an agent, running on a Notes 9 client, those two innocent lines of code gave me this error:
java.lang.NullPointerException at sun.net.www.http.HttpClient.parseHTTPHeader(HttpClient.java:738) at sun.net.www.http.HttpClient.parseHTTP(HttpClient.java:665) at COM.ibm.JEmpower.applet.http.HttpURLConnection.getInputStream(HttpURLConnection.java:411) at COM.ibm.JEmpower.applet.http.HttpURLConnection.getHeaderField(HttpURLConnection.java:703) at java.net.HttpURLConnection.getResponseCode(HttpURLConnection.java:399) at JavaAgent.NotesMain(JavaAgent.java:18) at lotus.domino.AgentBase.runNotes(Unknown Source) at lotus.domino.NotesThread.run(Unknown Source)
The connection worked, but getting the InputStream failed. That was especially weird because (A) I've used that code plenty of times before without error, and (B) it worked just fine on at least one of my other test clients.
Looking this up on Google I found only sparse reports of other people experiencing this: a question on StackOverflow and a thread on the atnotes.de forum. At the bottom of the atnotes thread there was a link to an IBM APAR that mentions this problem in Notes 8.5.3, but it was specific to applets... and it was opened in 2013 and closed about a year later (I have no idea if there was a fix involved, but it is marked as closed).
If this was a widespread problem there would have been a lot more chatter about it on the Internet, so I assumed that there was something specific about my setup that was causing the error.
After many restarts and toggled settings, and much pulling of hair, I was able to find a set of steps that would either cause the problem or prevent it. To make the code work:
To get the error:
There's something about opening the Java Debug Console before running the agent that causes Notes to use a
COM.ibm.JEmpower.applet.http.HttpURLConnection to make the connection instead of a
sun.net.www.protocol.http.HttpURLConnection, and the JEmpower version of the class is the one that's broken (per experimentation and the APAR mentioned previously).
Oddly, if I run the agent and then open the debug console and then run the agent again, everything is fine. It's opening the debug console before I do anything that causes problems.
My best guess is that this started with Java JVM 1.6.0 SR12, which was included with IBM Notes 8.5.3 FP3 and Notes 9.0.0, but it could have been before that too. In any case, now that I know not to open the debug console first, that'll be my workaround.
Here's a screenshot from earlier this morning:
The entries in blue are currently outside of business hours ("night-time" for my purposes), and the clocks update throughout the day so I can always glance at the times and say things like, "Oops, it's too late to call" or "They're probably at lunch" or "It would be so awesome to live in Hawaii".
Tested on Mac and Windows, IBM Notes client version 8.0.2 and up. It's been very useful for me, and hopefully it'll be useful for you too.
It's gonna be a great show. The session on that list is one I submitted with Kathy Brown as a co-speaker, where we'll be taking the fear out of Java programming on IBM Domino with concise, easy-to-understand (and useful!) examples. It's a brand new session, and we've got lots of ideas for code as well as integration topics.
Hurry up and register if you haven't already. I hope to see you there!
The easiest way to keep up is to subscribe to the monthly newsletter. To do this:
A link in one of your old blog entries doesn't work. Will you fix it?
Okay, I suppose it would be polite to explain. I apologize for being snippy. Let's say "probably not".
Content on the Internet is transient. This blog has been around for just over 12 years, and some of the things I linked to in the past just aren't there anymore, or they have been moved somewhere else. That's just how it goes. For me to constantly track, look up, and try to fix those things... well, it's just not that important.
Besides, you can use Google just as well as I can. ;-)
In the special case of IBM Technotes that have seemingly vanished (for example, Technote 1305829 describing how to construct an extension.xml file for Widget deployment), I really do wish I had a hidden cache of archived knowledge that I can share with you, but I don't. I'm sorry. I am getting into the habit of saving useful Technotes as PDFs in case they disappear, and I suggest you do too.
Do you have new links to your old LotusUserGroup articles?
Unfortunately I do not. When the old LotusUserGroup.org website was retired in favor of SocialBizUG.org, none of my old Clippings content got moved over to the new site. A few of my old articles have been republished at my request, but the SocialBizUG folks have told me that they're not planning to reprint my old stuff en masse.
I can't just reprint the old articles here either, because SocialBizUG more or less owns the content. I'm sorry and if that changes, I will let you know.
Where are the slides from the presentation you gave at ____________?
I do have a Presentations page here on the site, but it is woefully untended. I apologize for that too, and I also apologize for apologizing so much in this blog entry.
I will try to update that page soon because I think it's important, but in the meantime here are some presentations I've given most recently:
Did you stop blogging/podcasting? Can I have your stuff?
It's hard to say that you've ever "stopped" doing a thing, so much as you start doing it a lot less frequently. That's how I feel about blogging here or podcasting with Bruce. Sometimes you take a break. Maybe you come back to it later, maybe you don't. No big deal.
In the meantime, I'm still doing presentations and writing bi-monthly developer articles for Clippings, so that's something. And honestly, the stuff I do daily working on panagenda MarvelClient (sorry for the blatant plug, but that's what I do full time and it's really cool stuff) is either very esoteric or very product-centric, so it's hard to pull bits of that knowledge out and talk about it here.
That's all for now. Nice to see you again.
It turned 20 years old today. I know that because I emailed the serial number to Maglite support and asked them when it was made. They look up the number in their database and give you the date of manufacture. 11/19/1993. That's my big guy's birthdate.
A few things about this. First, I don't have any other electronic devices anywhere near this old that are still in regular use. I do have some clothes that old that might get worn occasionally (don't judge), probably some tools, definitely some books, but nothing that plugs into the wall or uses batteries. I turned on the Maglite as recently as a few days ago.
TRIVIA: I still used the original bulb up until this year. The bulb lasted 19 years. Even more impressive, vowe suggested that I check the endcap for a spare when it burned out. Spare was there, and it works! After all this time I don't even have to buy a new bulb.
Second, I continue to use it despite the fact that I've got a few other flashlights that are much brighter. I have a little $4 Chinese flashlight that I use sometimes (when I can find it) that puts out a lot more light, but the old Maglite is always the first thing I go for. I've dropped it and gotten it wet and probably even used it as a hammer; it has kind of an Old Reliable feel to it. It would be part of my Zombie Apocalypse Kit if I had one.
Third, it has that intangible brand loyalty thing going on. If this thing ever [somehow] breaks, I will go right to the store and buy another one. Maybe two (cause I kinda like the MAG-TAC line). I'll probably give one to each of my kids when they move out of the house someday. And then they'll be like "Really Dad, a flashlight?" and I'll be all "It's not a flashlight, it's an heirloom!" and they'll talk about it in counseling 5 years later. Whatever, I'm doing it anyway.
It mostly comes down to quality. I can buy cheap throwaway flashlights that are brighter and newer, but they're just... disappointing. This thing has good heft, it takes a beating, and it lasts forever (in flashlight years). I'll always pay extra for that.
Any comparison or metaphor to software, technology, or life in general is an exercise left to the reader.
But, like most things, it's good to delve a little deeper into the topic so you can really understand what's going on. Here are the IBM suggested settings and what they mean:
|Maximum Java heap size (memory used by the JVM) is 1024MB. The lowercase "m" at the end is quite important; don't leave it off or make it capital.|
|Initial (and minimum) Java heap size is 512MB.|
|Block size used for increasing memory allocation is 512kB (note that this value is in kilobytes, not megabytes)|
The technote includes some helpful additional information:
Xmcasetting is a block size for both adding and freeing memory, and is significantly higher than the default. This is as important as the
Xmxsettings but many people forget to set it. Don't forget this one! To quote the technote: "With the original settings, Designer can spend a great deal of time adding memory in small 8 K blocks or trying to free preassigned memory blocks when the 256 MB upper limit is reached. As a result, it appears to hang/freeze when actually it is working to allocate or free up memory in order for the program to run."
So that's good stuff right there. Maybe all you need to know. If you want to see some screenshots to back up the instructions in the technote, the TLCC page on this subject is a handy reference. If you want a bit more detail, please keep reading below.
First, 1024m is a very good setting for the max heap, even if you have tons of RAM installed on your machine. This is because the Lotus Notes client on Windows (at least as of version 9.0) still uses a 32-bit JVM. You might be running Windows 7 64-bit with 8GB of RAM, but you've still got a 32-bit JVM. According to the IBM Java documentation, the maximum amount of memory available to a 32-bit JVM is 1.8GB. However, some people have problems setting the memory too much higher than 1024m -- it supposedly has something to do with contiguous memory allocation, but I really don't understand the details -- so 1024m is a safe value. If you have less than 2GB of RAM on your workstation (and I'm really sorry if you do), don't go higher than 50% of installed RAM.
Second, you should set the minimum heap size to be smaller than the maximum heap size (as recommended). There are plenty of web sites that will tell you (especially for Java servers) to set the minimum and maximum to the same value for efficiency. Unfortunately, for a client application like Lotus Notes this can potentially be less efficient. Not only can setting the values equal prevent effective garbage collection, it can also cause heap fragmentation. This document on IBM JVM heap sizing has a very nice explanation.
Third, these JVM settings are normally placed in the
jvm.properties file in the Notes framework directory. If you install a FixPack or a new version of the Notes client, your custom JVM settings might get overwritten. Some (many? all?) FixPacks and upgrades reset the jvm.properties file. Every time you upgrade or patch your client, you should re-check the
jvm.properties file and reset it if necessary.
Fourth, you can also put these settings in the
rcpinstall.properties file. This file is in your
data\workspace\.config directory, which has the unique advantage of being editable even if your Windows setup has the Notes program directory locked down. It also overrides any values in the
jvm.properties file. Keep in mind that this file might also get overwritten when you upgrade your Notes client -- or if you delete your workspace directory.
I personally prefer to put the values in the
rcpinstall.properties file, especially because this has priority over the
jvm.properties file so I have much more confidence that the settings will be used. If you're curious about what the final values used by your Notes client are, you can go to the menu option Help - About IBM Notes and click the "Configuration Details" button, then scroll down to the
Or, as mentioned in the technote, you can open Domino Designer and use the "Show heap status" preference. A couple things about that preference though: (1) it only works for me if I unset and reset it after I restart Notes, and (2) you need to hover over it to see the actual max heap size, which is labelled as the "Mark" size on my Notes 9 client. The "max" size listed in the heap status bar is usually less than the max heap size.
Next, I am happy to announce that Chris Miller wins the "Guess The Mud Run Time" prize. He will be receiving a very cool piece of art by Matt Reimer himself: a drawing depicting an inhabitant of Matthewland. Thank you Matt and Troy for offering that as a prize.
If you are morbidly curious about what it looks like for me to participate in a mud run, I have put together a special mud run tips page with bits of advice for anyone else who might want to try this sort of thing. At the bottom of the page is a link to a gallery of pictures that my wife was fantastic enough to take as she tracked me (with a huge grin on her face) at various obstacles.
If you're wondering: yes, I did get pretty dirty. But my face stayed clean and the mud stayed out of my mouth. So that's something.
Finally, for your viewing pleasure:
In the actual video you can hear my daughter yelling "Lie down in the mud Dad!". Children are so precious.
Upon recovering from yesterday's mud-in-the-face bit, I decided it would be better to hold off on any more intentional dirtiness before the race. I have switched metaphorical tactics from "wading slowly into the pool" to "jumping off the diving board". Tomorrow will be the dive.
So today I have been picking out clothes for the race. Here's what I came up with so far:
The rationale is:
That's right, I showed you a picture and offered a rationale for my running undershorts. Please donate so we won't have to do this again.
Thank you for your support.
I apologize for the slightly large size of the animated gif (I usually try to keep graphics much smaller than that). I just thought it might have a bit more impact that way. Plus, animated gifs are all the rage this year, aren't they?
The kids originally thought that they were allowed to throw the entire bucket of mud on my face... I had to explain to them (twice) that they were allowed only a handful each. I was glad for a nice shower afterwards.
As part of my last-minute training for this event, I will practice getting dirty a little more and a little more each day. Here is day 1:
Hey, you gotta start somewhere. Georgia red clay is very... mushy.
It is also important to set proper expectations. I knew that a mud run would involve mud and general dirtiness, but I'm not sure I appreciated quite how much filth is involved until I looked at a few "What to expect in a mud run" articles last night.
There are frightening stories comparing post-mud-pit to being an unwiped baby, and warning runners who don't normally wear underwear (really?) to buy some nice long spandex. Apparently you have to be very careful not get your shoes sucked off in the pit too -- some folks use duct tape to secure them to their legs -- and people advocate taking a running belly flop leap into the deep mud rather than going feet first.
Some of the important points seem to be:
So it seems that I should wear spandex, rubber shoes, and gloves. If I add a cape I'll be a superhero. I'm considering shin guards and a surgical mask too. Maybe goggles.
Longer version: I know that a lot of people have been asking for donations for various charities recently -- people in the "formerly known as the Lotus Community" and elsewhere -- but please take just a minute to read this.
Troy Reimer is a good friend of mine. I worked with him at SNAPPS a few years ago, he's an excellent programmer, and he's a really really nice guy. You might have met him before, or seen him speak at a conference, or used some of his code. Even if you don't know him personally, you're only one degree of separation away from him as you're reading this.
Two years ago, Troy's 10 year old son Matt was diagnosed with a genetic disorder called Adrenoleukodystrophy (ALD). Click through to the About ALD page on the Art of Curing Matt website for details on what ALD is and what it does to those who have it. Because it's a relatively rare disease, there is a very limited amount of research into treatments and potential cures.
I won't be at the art auction this weekend, but I wanted to try to do something to help raise money and awareness.
On the same day as the art auction, there is a mud run near where I live. People who know me also know that I don't like dirt. Or getting dirty. Especially not intentionally. However, I will allow myself to get covered in mud and other outdoor woodsy filth for a cause. So here's what I'd like you to do:
I don't really have any good way to track the money that I'm personally raising -- I'd like everyone to donate directly instead of going through another site that would charge administrative fees -- but I will maintain the list of guesses about my race times. Hopefully that will result in some small amount of competitive activity to keep everyone engaged for a few days.
I will also post pictures this week during my "training" as well as after the race, for the curious.
p.s. - yes, this is very last-minute. But every extra dollar you donate is a dollar that the researchers would not have had otherwise.
If you are using OpenLog with LotusScript on a Domino server, please set
This is a global variable that determines whether or not the OpenLog code will use Lsi_info(14) to log LotusScript stack traces. IBM has found that the Lsi_Info function (which is part of the LotusScript language) can cause serious problems on the Domino server. You REALLY need to avoid calling this function on the server.
The good news is that (A) as far as I know it is still okay to use this on the Notes client, just not the server, and (B) if you are using OpenLog it is very easy to turn it off globally or on an agent-by-agent basis simply by setting
NoLSIStackTrace=True. The bad news is, well, you lose debugging info. But the good far outweighs the bad.
DO NOT PANIC. Just go into your OpenLog database, open the Initialize sub in the "OpenLog Functions" script library, and set the NoLSIStackTrace variable there. Then save and close and use the "Design" server task to push it out to all your databases that use OpenLog. There, you're done.
Don't get scared into thinking that you have to stop using OpenLog, just set the variable and push it out. That's the beauty of deploying script libraries using master templates.
John Hodgman was interviewing Steven Strogatz about mathematics and his book The Joy Of X, and the discussion came around to the statement that the number one is actually equal to the number 0.99999... repeating infinitely. Steven Strogatz asked if anyone in the audience had a "favorite proof" of this and, well, I indeed have one.
So I got to stand up, find a microphone, and stumble quickly through the proof. I was amazingly nervous -- being on a mic in a room with a few thousand people explaining a math proof to a well-known mathematician will make you nervous, as it turns out -- and I feel like I wasn't very clear about the whole thing, but John Hodgman made some very funny "WHA?" noises and both men on stage seemed to understand what I was saying and roll with it. And people clapped. So all in all it was a win.
The proof goes like this:
I've always thought that was a neat equation, and quite frankly it's about all I remember from my infinite series and sequences class in college. Math is cool.