I was very saddened to hear that Arthur C Clarke had died. I grew up with his books and enjoyed their resonance with my love of science and technology.
Here is his ‘good-bye’ message, recorded recently around his 90th birthday. I’m impressed, as ever, with his prescience, though I wish he’d been around to record another on his 100th.
I have just had a call with IceSave, with whom I have opened a new savings account. I have been consistently unable to log in and so wanted to work through the error. The nice girl on the phone confidently told me that all she needed to do was ‘refresh the internet’ for me and it would work tomorrow.
Technology’s amazing.
Update 7th Feb: Incredible. It works. Next time you’re suffering login problems, I suggest you try refreshing the internet. It worked for me.
Look at this article in The Register. It gives a perfectly plausible explanation for the alleged Iranian threats to the US navy in the Gulf. If the US were not trying to drum up public opinion against Iran, leading most probably to another conflict, then surely they would not have gone to the press with an accusation that stands on such wobbly pillars of evidence.
ProClarity Analytics Server (PAS) is a web site providing storage of OLAP ‘report’ definitions and caching of the resulting reports. The ‘reports’ are organised by a hierarchy of libraries, briefing books and views. The metadata for all of this is stored in a database.
In order to copy these libraries, books and views between servers, ProClarity/Microsoft provide a feature in a tool called PAS Administration. There is a "Migrate Repository" button that leads to a wizard where a source and target database are specified. On the last page of this wizard is a "Migrate" button to begin the work.
Typically the migration should take under a minute. However, after performing the migration many times we were finding the migration took longer and longer – up to six hours in the end!
My colleague Dan Meacham ran profiler to check what it was up to. It turns out there’s a table called SecurityTimeStamp against which it issues a "Select *…". I’m not sure what causes insertions into this table, but on the various PAS servers we checked there were between 70,000 and 1,200,000 rows in this table!
A swift "TRUNCATE TABLE SecurityTimeStamp" on the source database brings the migration time back down to a few seconds again.
We’re using version 6.2. Version 6.3 exists and may fix the issue without resorting to dodgy database hacks. Nevertheless, I hope this saves other people a bit of time.
PowerPoint prints hidden slides by default.
Tell me, why am I preparing training material? Not quite in the job description of a developer…
A couple of days ago I got off the bus slightly behind a South African couple. A short distance away, near the entrace to Sainsbury’s, stood an elderly eastern European lady selling the Big Issue. I realised the South African girl must have only just arrived in the UK at the point that she decided the Big Issue was a free handout and attempted to liberate one without breaking her step. Her boyfriend was in fits of laughter and declared his intent to tell all her friends that she tried to steal off a Big Issue seller.
I was amused, but the old lady less so as she never did make the sale.
I’m going to focus on some aspects Calculations in Analysis Services cubes here, and perhaps follow up with more areas later.
1) I’ve already mentioned the strange location for setting the Display Folder property of Calculations in another post, but it deserves inclusion here. Additionally, why is it not possible to set other properties such as Description? Why can we not use the standard Properties window in the shell for all these? Why are Format String, Visible and Non Empty Behavior (sic) on the made expression editing page. (Actually it’s because the Display Folder is in the <Calculation Properties> element and the latter form part of the MdxScript, but this is not a good reason for the UI to split them up.)
2) Intellisense in the Expression box, and real time evaluation of member names. The real time Expression validation in the calculations only deals with basic MDX syntax. I would like a background thread to check member names as well. I accept that it would be difficult to differentiate between the intent to reference a data driven member versus something like a measure name that is baked into the schema. However, it would be useful to at least have mis-spelled measure names underlined. Errors in these are only caught when the calculation is first queried.
3) The Metadata pane in the Calculations tab does not show calculations that have a Visible property of false. This is pretty irritating when you try to drag and drop invisible calculations into the Expression box. It encourages error prone re-typing.
4) The "Calculation Properties" window is too narrow on opening. With reasonable sized calculation names with common prefixes it’s impossible to tell them apart due to the short column width.
5) Comments in calculation expressions cannot be placed at the top of the calculation. If you switch to script view and back to form view they will no longer be visible (but they are still available in the script). They work if placed at the bottom.
Lately I stumbled upon an email signature in which the author had entitled their job "Principle Engineer". As it has never been explicitly clarified, I suspect this is a homonym of the intended adjective. That is, unless his job was to maintain the moral upstanding of the company.
Of course, a "Principal Engineer" ought to be the first in order of importance [Cambridge Dictionaries Online], which is also not true as there are, possibly, hundreds of employees in his company with this designation.
There is no Microsoft.Jet.OLEDB.4.0 provider for 64 bit Windows editions. If you need to use an Excel Source task or Excel Connection Manager in Integration Services on a 64 bit machine (I’m using Vista Ultimate x64), you will need to set your packages, or your project, not to use the 64 bit runtime. This is done by setting the Run64BitRuntime property to false on the project properties’ Debugging tab.
This article on MSDN provides more information about Integration Services running 64 bit editions of SQL Server and Windows.
Also note that if your source file is Excle 2007, the Excel Connection Manager cannot be used and the OLE DB Connection Manager must be used instead. See the note in the linked MSDN article about the Excel Connection Manager.
Technorati tags: SQL Server 2005
How do you perform a Lookup transformation using an input column of type DT_DATE (e.g. from a spreadsheet input) and a SQL Server table column that has the datetime type? Dragging the former onto the latter results in the error message "One or more columns do not have supported data types, or their data types do not match".
The answer is to create a Data Conversion transformation earlier in the flow. In this task, convert your input column to the type "database timestamp [DT_DBTIMESTAMP]". Make a note of the output alias you have entered. Now go back to the Lookup transformation and drag the new column (i.e. <output alias> of the Data Conversion transformation) onto the datetime column in the lookup table.
Josh's Space 