Today I had VB6 crash with an error reported in the VBA.DLL module when loading the project. Reloading VB6 had no such error which is a bit fun. I’ve finished completing the reuploader tool and its serving its temporary purpose and apparently working properly.

I also had some ePrints performance fun, however its now five minutes later. I still think its a cache expiry issue, however I’m not sure yet. The query itself doesn’t appear to come from ePrints but some PHP file somewhere, so I’m not sure what is its trigger. There were also some minor ePrints changes, mostly cosmetic alterations to the way things looks and links. I’m scheduling these changes for release on Wednesday, it shouldn’t result in any downtime so we can do it during the middle of the day with enough precaution.

Today I also had induction. I’m sure it would be useful if I hadn’t been studying at the University for the last few years, hadn’t of already started working and if it wasn’t the regurgitation of the web site and the contents of it. Still, it served to kill an hour and a half of time.

Another day of Visual Basic 6 fun. The VB error of the day is below:

This charming little error occurred reproducibly when I try to load up two copies of VB6 at the same time. I have two copies running with the new project I’m working on and the older project that I’m trying to pull code from to improve. Dismissing the dialogue doesn’t appear to cause VB6 to crash horrifically so I’m not quite sure what it is about but its a wonderful error though.

Developing more and more with Visual Basic and hitting its limitations and today yay for Microsoft: http://www.thevbzone.com/secrets.htm – secret API functions! You know, the ones that Microsoft said didn’t exist and said didn’t give them a competitive advantage? Yeah, look at that, they are there! At least I have callbacks but in the end I worked out a way to solve the problem differently that avoided the need for callbacks after fighting with trying to get them to work. No fun, but anyway.

For the most part I’ve got the backup system (of sorts) mostly functional and working. It doesn’t shell out to handle stuff and has some better error reporting but its still not perfect. I guess I can do that on Monday. At the moment it should be enough to get us across the line and thats an improvement on where we are now.

We also had some ePrints weird slowness. Personally I think its the system trying to regenerate itself from the cached data because its expired. It then doesn’t help that they think its slow and then start clicking around the place to help make it even more slower. “It feels slow, I’m clicking on all of these links that I wouldn’t normally click – hey! Do you think that has anything to do with it?” Gotta love users.

So today Windows was (unsuprisingly) slow again. It takes 20 minutes to get not very far when my computer starts in the morning and it makes me wonder if the login process subtly lies to you. The other day I started my computer but disappeared to help someone else out. When I came back and logged into my computer it was almost lightening fast to get to the desktop. Today when I do it it states its applying settings, running login scripts and a few other useless messages before actually logging me in. Is Windows trying to fake its boot time by prematurely loading its user login dialogue even though its still loading in the background? Rather annoying.

I’m still ahving fun relearning visual basic and its weirder limitations with ‘components’. I keep wishing to use a ‘component’ completely programatically however it appears the easiest way to do this is to assign them to a form somewhere. Perhaps I’ll work out how to get at them entirely programmatically but perhaps that just isn’t possible.

Following on that note I’ve found that the Microsoft INET control that they offer is completely pathetic and useless so I’ve taken to integrating curl into the system to handle the file upload aspect of the system. It seems a whole lot easier and logical that constructing entirely the body and headers myself and then telling the inet control to handle it – might as well just open a socket connection for the extra few headers I’d need to construct. I’m also looking at integrating lame with VB6 as well to handle mp3 conversion for the other application that I’m going to need to rebuild sooner or later. All in all much fun integrating various external pieces of technology together. The interesting thing about curl is its use of some of the more lower level functions that Microsoft offers as hidden functionality and is exposed by a specific API. IMHO its very interesting and facilitates some nifty toys within Visual Basic but at that point I’d almost be tempted to just cut and run going to a better language.

Today started with a prime VB6 annoyance: the project deciding that it didn’t want to work and wouldn’t load:

It turns out that this is a known error thanks to our friends at Microsoft with IE7: http://social.msdn.microsoft.com/Forums/en-US/iewebdevelopment/thread/21935021-1dc7-445b-a829-b02489009aab/. I had installed IE7 the other day but the bizarre thing that gets me is that the project worked perfectly fine the day before and then suddenly became broken for some reason. I’m not quite sure how that happens since the IE7 installation took place a week ago but that forum thread had the solution to my problem. I used the solution where in you change the reference instead of the registry hack, not sure which should have been better but it worked.

The next bit of fun I had was remembering how to quit a VB6 application. For some reason the application wasn’t quitting properly because I had forgotten to unload a form somewhere. After all of this time I’ve been used to calling a function that killed the system or halted the run loop so this was a bit of a spin for me again to have application existence controlled by a window. VB6 tries very hard to tie you to a window (or “form”) somewhere to the point that some items don’t appear to be programmatically accessible unless they exist as a ‘control’ on a form. Sucks but thats life.

With all good dinosaurs, they have issues with the modern world. The VB6 IDE is very much a dinosaur and has no mouse scroll wheel support. Thankfully someone at Microsoft realised that having the IntelliPoint software wasn’t a real fix and released an add-in that does the trick: http://support.microsoft.com/kb/837910. Add-in’s in Microsoft Office and its IDE tools appear to be one of the better thought out and developed aspects of the system. The ability to extend the functionality easily I feel has been done well. I might bash Microsoft a lot for their bad stuff but there are tonnes of items that they do well too – and add-ins at a macro level I think are one of them.

Another annoyance I had today was how VB6 has a habit of dislodging files from project repeatedly and when you restart it wants save file wrong directory. Instead of automatically selecting the project file’s directory when available, it always defaults back to the program file directory instead. So you end up having to wander around the FS to get back to your project if you want to save a file after you start the IDE when continuing on an existing project. After you’ve saved one file it’ll remember the directory for the rest of the session but once you shut down the IDE and jump in again you’re back to scratch.

I’m also relearning how classes work in VB6, its been a rather long time since I’ve done work in it so I’m back to picking up things. Some stuff sticks around like using “Option Explicit” to enforce explicit variable declaration instead of the default implicit declaration on use, however some other stuff has disappeared. One of which is the use of classes within Visual Basic. Compared to almost every other language its support is beyond limited but it is still at least there I guess – so I figure I should sort of be grateful for at least that. However its still a bit sad that it doesn’t have more full support but I guess it is a bit much to ask from a system a decade old (mind you C++ is over a decade old in various incarnation and had support for objects, as did Java and I think even Python at the time but anyway).

To end with on a Joomla note, the site of the day is the Botswana government site (http://www.gov.bw). It looks like a Joomla! 1.0 site and I saw it when one of the Uni’s email’s to the domain bounced and I wondered what they did. Very cool.

This morning we had a presentation on the Oracle IDM system. The ‘IDM’ is comprised of OIM (Oracle Internet Manager?) which pushes (at least in this system) information into the OID (Oracle Internet Directory; a directory which exposes an LDAPv3 interface). The system has a rather hackish, ugly user edit screen, generic technology converter, limited AD integration (push only, no pull) and the system appears to have limitations on its fields, for example it can’t hide fields on new user screen as opposed to the edit screen. Bummer. They’re purchasing another product to handle password changes from Active Directory but thats rather weird. Coming from an environment with Novell, it seems strange that it can’t handle it but anyway the University has bought all of the Oracle junk and now it’ll be time to put it back together. The team I work in (Learning and Teaching within Application Support and Development) has the sneaking suspicion it will come back to us, either our group specifically (L&T) or our parent group (AS&D). Either way we’re not looking forward to getting the mess.

Today also revealed a queer ePrints security problem. One of my deployment scripts failed to reset the security on the external site which caused the system to be inaccessible externally. The change was made on Friday morning so it took until this morning (thats 4 days later) to detect the issue. The question was asked how much attention is really paid to the system, but thats neither here nor there. I fixed up my mistake and reran the server configuration scripts and it was all back and running. Whilst I was working around with it I found some large ePrints log files – over 2.7GB of log files in the archives going all the way back to 2007 and none of them compressed. Having a chat with people we’re probably going to burn them off for reference and reclaim a bit of space.

We had some Moodle fun with one of my issues appearing to be a mysterious self fixing problem (could do with more of those). We’ve had other changes that impacted on that area in part so perhaps that fixed it. There was also a Moodle migration today and I believe one of my changes are now in production – already I’m leaving my mark on the systems!

Last but not least, the Adobe Presenter is broken after just two days of semester. In those two days however its managed to generate 400MB of data. This happened late on the day so the rest of the week will be taken up fixing up this stuff as well as shifting data when we get a larger disk to store the presenter stuff.

Today was a bit of a mixed bag and I started off with some ePrints minor customisation work. Nothing major so far, but that is certain on the horizon, its just cosmetic changes more than anything.

I also had some more Moodle fun configuring my box to connect to the internal Moodle Oracle Development server. Looking through the config file I came across this at the bottom:

// MAKE SURE WHEN YOU EDIT THIS FILE THAT THERE ARE NO SPACES, BLANK LINES,
// RETURNS, OR ANYTHING ELSE AFTER THE TWO CHARACTERS ON THE NEXT LINE.
?>

Basically instead of just not putting the final close for PHP, they put it in with a warning. I used to be in the habit of doing this however increasingly I’ve stopped putting it at the end of files for the simple fact it isn’t required and using it can cause issues, which is what the above is causing. I’ve also discovered that Moodle has some serious memory issues: it can use up to 15MB to render a simple page – this is one hungry application!

Yesterday I tried to grab some money from my bank account only to find that my NAB debit card had expired. This is weird because I didn’t think I had received a new card. Tripping off to the bank today during lunch time revealed that due to the bush fires in Victoria the bank hadn’t sent me the card. Whilst I was there I also double checked that they had the right address on file which they do. They’ve extended this morning and believe that I should get card this week.

This morning started again with another ePrints deployment at 8am. This is the third week running, one per week, that I’ve done an ePrints deployment of some variety. This morning was a run to get SVN back in sync entirely, so now production is fully aligned with our test and development environments. There was still one outstanding issue, but we’ll address that as we go and its a minor change so we can do it without taking the service offline. Later on in the day I also wrote some ePrints migration doco so we’ll see how we go.

The Outlook look annoyance of today is that the “Delete items on exit” feature works only once per login. After you’ve quit for the first time, quitting again doesn’t seem to retrigger it. Logging out and then back in again causes it to redisplay the query to delete it but again, only once. Rather annoying.

I had some fun on Moodle again today, learning more and more about the system and how it goes together. Will have to do some more again next week, fun with one of the clean up scripts appearing to be a bit excessive, but that shouldn’t be too hard to look at.

After lunch it was a tad strange walking into the building and hearing the crickets almost as if I was outside. Some how the building I work in seems to be a habitat for crickets, so I’m not quite sure what is causing it but even on level three we have crickets appearing on the floors. Weird!

We had a meeting about how we structure our various Subversion repositories. Where possible we’re going to use SVN externals to build our Moodle tree instead of having everything live inside the Moodle directory. This will mean USQ developed extensions will live in their own repositories managed by themselves and then the core of Moodle will form the basis of the tree. This will permit us a lot of flexibility in the long run but we’ll see how we go with it. The one down side of this method is that it won’t support individual files at all and even in 1.6 not too well. We’ll see how things look and then progress from there.

Directly after our Subversion planning meeting I had fun with the ICE crew learning about ICE ORE, SWORD, ePrints and how the ICE team has them working together. They’re rather hard line about their view of the world and that the web should entirely be built from web accessible documents and want the end of PDF. Their demo included pushing a HTML document only and hiding the PDF of the actual paper beneath their HTML document. They didn’t seem interested in actually doing any statistics gathering on their suppositions either, which does make me worry about the team. Personally I would have thought they would have loved the chance to prove themselves correct, but they seemed to avoid wanting to gather information about user directions – what I’d really like to do is set up some A/B testing with a page displayed using their preferred method (HTML is the only option and you have to click on it to get the PDF) against HTML and PDF together. Implementation looks straight forwards so I’ll push it into dev and from there I’ll let the politics sort the battle out.

Another week over.

Another day, another useless Outlook message. The message of the day for Outlook is “Done checking your data file”. As a user I really need to see that message with a little warning logo, personally I wondered for a second it it had been corrupted, why else bother me? But then I remember something I read once comparing Windows to Mac: Windows is the impetulant child that always wants your attention whilst Mac is the butler tries to get out of your way. I don’t have annoying messages like this on my Mac, even if Mail is perhaps doing tasks on my “data file”. RIL is annoying me again compared to Quicksilver, items aren’t prioritised and useless entries (like DLL files and HTML files I’ve never looked at before) are in list ahead of the application I really want to look at, which was the MySQL Query Browser at that point. I get around it by putting ‘exe’ at the end as I type and this usually reduces the resultset to just what I want but its still a lot worse than Quicksilver. What should I expect from a Windows tool? On the subject of annoyances and Windows, Outlook for some reason has a particular hatred of the ePrints tech mailing list. For any HTML email that comes through it doesn’t open it but decides to display the signature of the list and gives me the option to double click on it. Strangely it doesn’t think it has a document previewer for it and when I double click the file opens fine in Chrome. A mild annoyance I must admit and I’m not entirely sure if its Outlook’s fault entirely though the lack of a ‘HTML’ file previewer is a tad depressing.

I also had Moodle character fun with restores not uploading properly due to legal UTF-8 characters and what I presume is PHP’s XML parser library though I’m yet to dig into it. It shouldn’t be causing an issue but it appears to be, for the time being the work around is not to use those UTF-8 characters (in this case it was a right single quote). In other Moodle news, I wrote a new group export tool for the Engineering department’s problem solving unit. It exports in tab delimited like they requested and maybe one day I’ll get around to making the XML file format that they wanted as a preference. However, walk before you run.

Still Excel and Sharepoint forget it’s password after the Office 2007 upgrade, it is annoying to have to put my password in to get to my time sheet. I’m told by a Windows fanatic friend of mine how wonderful Microsoft is in a pure Microsoft environment – but to be honest, its looking like a sham. I’ve got Windows, its joined to Active Directory, they have Sharepoint running on IIS boxes and I’ve got Office 2007 and it still doesn’t automatically authenticate me properly. Yes, life in the Microsoft world is always so much better, can you just re-enter your password again? Thanks.

So today my Outlook decided to be horrible non-responsive as it died under the deluge of rejection emails that the USQ Studydesk email received. I had fought with Outlook to set up a filter for the mailer daemon previously and had decided that given that the amount of messages wasn’t too bad I’d push it into a folder. How I was wrong. So now I decided that I’d immediately delete it. Upon looking at the Outlook 2007 Rules configuration a strange new “recommended rule” had appeared. I did a quick search and it looks like it can be quite nasty if I tried to downgrade from Outlook 07. Fortunately I’m still on Outlook 07, so I removed the rule. So I decided to alter the existing rule to delete hte messages. After a short while watching my deleted items autonomously rise I resolved to instead just to kill them without review.

Some how it decides that I must be running Outlook before it will automatically kill the message, however it will happily just delete the item (equivalent of moving) without user intervention. I must admit this differentiation between “server” rules and “client” rules is baffling coming from a Notes background. Notes, for all that is horrible about it, was in some respects really well designed. It wasn’t much to look at but it offered some really awesome features such as great document replication, offline modes that worked well and the fact that “applications” could run on the client or the server, so your rules always worked. This turned out to be really awesome for when I moved from a Windows box at TCC to a Linux box and found that all of my old email filtering rules continued to work perfectly fine even though I was working via IMAP at that point. Very cool in my opinion. Unfortunate that Exchange doesn’t have this ability and in fact even more of a shame that IBM sat on their heels and let it go to waste.

In other news I tried out the Safari 4 Beta today. The “Top Sites” feature appears broken for some reason appears ‘black’ on my computer. It sort of worked upon first install and its great to see Safari stepping out already and using a video tag for its intro page. It has in part some aspects of Google Chrome with top mounted tabs. I haven’t had much of a change to use it but it has a few changes. I’ve downloaded and installed it on my computer at home but I’m going to need to restart my Mac first. Funnily enough it saved the day for the Mahara/load balancer fun we have been having for the last few days.

For the last day or so the Mahara team has had fun trying to get the test system to deploy fine. After success with the development environment, the test environment with the load balancer raised many issues. IE returned nothing useful beyond that it couldn’t hit the page and Firefox returned a redirect loop error. I tried to use Firebug to see what was happening but couldn’t see it as Firebug seemed to temporarily work and then once Firefox determined it was in a redirect loop it killed Firebug’s output. At one point blame was laid at the load balancer, much time was spent reconfiguring the Apache server to get it to work including misconfiguring it and reconfiguring it. Eventually Safari was slow enough to display the redirects to figure out that the load balancer was redirecting to HTTPS whilst the Mahara was then redirecting back to HTTP. After quickly reconfiguring Mahara it appears that most things are working properly. Unfortunately this then lead us to a problem with the MySQL database which was strangely reporting tables existing with “SHOW TABLES” however when one selected against it the MySQL server reported that the database didn’t exist. We ran through all of the GRANT’s to ensure that it looked fine however everything looked right and so late in the day we decided not to bother much and leave it for tomorrow.

Last but not least is a shout out to one of my more favourite windows tools, the RockIt Launcher (RIL). The tool is great though it is no comparison to Quicksilver, which is the Mac OS X tool that I was trying to replicate. RIL is a launcher on its own however Quicksilver has a bit more power. However, RIL meets the need for a keyboard based launcher tool for Windows and its much appreciated.

After having fun with the presenter project we managed to get everything mostly sorted today. Although the desktop deployment job failed last night they were able to get one of the lecture theatres up and running with the new build for testing. Thankfully this worked fine with everything working the way that it should with no crashing of PowerPoint to be had. Other issues have since cropped up on the student systems in addition to all of the lecture theatre computers. My part is done however, so we’ll see what happens on Thursday for the public demo.

I also managed to today fix up the dev environment of ePrints and wrote this down for future reference. It turned out that ePrints has a simple command to resync the database. At this point I managed to update the test environment to be in line with development. There have been continuing issues with the production environment having changes that weren’t made to either of the test or development environments. At this point I’m hopefully going to be able to get everything back up to date though that will be later in the week.