Archive for August, 2007

Web v Desktop

Friday, 31st August, 2007

I read an interesting article in PC Pro magazine the other day. here is roughly how it went (with my interpretation thrown in):

Flash has been the dominant force in rich web UIs for a long time, Adobe are now releasing Apollo (renamed AIR). The Adobe Integrated Runtime will allow flash based apps to run on the desktop. This is Adobe coming back from the web to the desktop. It could easily be the difference between the old Macromedia view and Adobe of course.

Meanwhile Microsoft have released Silverlight to much fanfare and hype. Silverlight is basically Microsoft Flash. This is Microsoft, master of the desktop moving to the web. Of course the whole .net thing is part of that too.

Just seems odd to me, both companies dominate their home turf, yet both seem determined to get a piece of the other market. Reminds me of the way the kids always want the toy that the other one is playing with. Grass is greener??

It also gives me some confidence that reports of the death of the desktop are exaggerated.

Is anyone else following this trend?



Multi developer Excel projects

Wednesday, 29th August, 2007

Has anyone else worked on a spreadsheet development project where more than one person was developing the spreadsheet part?

I know most of us will have integrated our work with developers of other components, here I’m thinking about you do 1 sheet (or range) they do another?

If so what was it like? how did you run things? did you use source control?

I did one project where there were 2 of us working on the grid part and the VBA. It was a real PITA.

We split up the worksheets and worked on different parts and re-integrated regularly. re-integration was something we dreaded, and over the months a few bits of work just sort of disappeared, or got overwritten with old versions.

We tried having one master copy and taking it in turns to work in that, and to work off-line. We still had to integrate, it was still a pain. We tried pair programming, but basically that was one person typing and one person being very bored. Pair programming works great for complex things, its pretty boring though for the non-driver when the work is fairly mundane.

We both agreed that either one of us could have done the project faster on our own. I would actively avoid this sort of work in the future, unless there was a clear workable way to break down the work so we wouldn’t be tripping over each other all the time.

The best part of our multideveloper spreadsheet project was when the other dev was on holiday and you knew you had the master all week.

The difficulty of multiple developers is one of the real drawbacks of spreadsheets in my mind. Breaking up the tasks to share them out and so reduce the elapsed calendar time is a real challenge because integration can be tough. Unless you know different? Maybe it was just our design? Or maybe the system was too focused and too interdependent?

What do you think?



Business tip

Monday, 27th August, 2007

Here is a business tip for any self employed folks out there.

There are some accounting rules that enable companies to take the cost of certain projects to the balance sheet. Normally costs just go straight through to the P&L as expenses and come off the profit in the year they are incurred.

If a project can meet the criteria to be capitalised then the costs can be spread over multiple years. I won’t go into the minutiae of those rules for 2 reasons:

  1. I don’t know them
  2. They are probably pretty complex
  3. If I did know them I wouldn’t bore you with them

ok 3 reasons then.

In capitalised projects the value that gets put on the balance sheet (as an asset) is related to the costs incurred. Therefore – the more they spend the more the asset is worth. Neat.

If you ever get chance to work on this type of project (they tend to be fairly big) there is often significantly more rate flexibility than your standard expensed project. After all the more you charge, the more their final asset is worth.

Of course the costs do hit the P&L eventually but normally depreciated over many years.

Anyone else worked on these sort of projects? Any other top tips?



Quiet week warning

Sunday, 26th August, 2007

I’m out an about most of this coming week 27 -31 August, so the post rate will drop I’m afraid. I should be able to manage a couple of posts but no promises and no comments probably.



Microsoft Office Developer Advisory Council

Thursday, 23rd August, 2007

A load of us are off the Seattle in a few weeks to discuss the next version of Office (official codename: Office 14 I think). This is a follow up event to the one I went to in March.

Its a great event where we look at whats coming up and give our views on how things might work. Its good to see MS including a wide variety of people in their thinking/prioritising. The specific focus of the program is on the developer story for Office. Its all under NDA of course, but I can certainly pass our views on to them.

So from an Office developer perspective,

  • whats good and needs continuing
  • what needs improving
  • what needs adding
  • what needs to just go away?

What I would like to see in OfficeNext is:

    1. Improved/updated Integrated Development Environment – either the VBAIDE (at least to VB6 standard) or preferably VSTA (with single file solutions as an option (ie not forced to split code and grid))
    2. Improved external data management
    3. Introduction of performant User Defined Functions.
    4. decent level of .net integration (preferably including distributing a recent framework)
    5. Improved compliance/auditing features, including decent workbook compare functionality.
    6. Fix the UI or bring back commandbars (same thing??).
    7. make range.dependents/precedents include off sheet ranges as well.
    8. Don’t know about collaboration features – what do you think?

      So thats me, what sort of stuff would you like to see?



      Mention on AccountingWeb

      Tuesday, 21st August, 2007

      We got a mention on AccountingWeb for the spreadsheet hell paper and the Eusprig presenation.


      The stuff mentioned can be found via this post

      Frankensheet got attributed to Rob Bruce ok, but versionitis should have been credited to Marcus.

      I really feel that the spreadsheet quality story is beginning to get recognised. One thing that will help that in my view is the message.

      Telling people their spreadsheets have errors does not work

      • a. they don’t believe you, unless you prove it on their actual work
      • and b. they dont want to hear that negative message.

      I think we need to change the message, and I think the message should be:

      ‘spreadsheets have dangers, let us help you steer a safe and a productive path’, or something in that vein, but perhaps more eloquent.

      I think that is starting to happen,  Steve Powells paper this year, where they suggested the biggest risk seemed to  be the wasted time and effort of using poor spreadsheets,  opened a few peoples eyes I think.

      Have you got any views about what the message should be to encourage an improvement in spreadsheet quality?



      Office 2007 beta service pack 1 rumoured.

      Monday, 20th August, 2007

      Sounds like an early beta of Office 2007 Sp1 has been released to some preview testers.

      Please please please let it have a UI compatibility option. Don’t hold your breath for a broad release though, O2007 was about a year in various betas I think.



      Excel – COM v XLL

      Monday, 20th August, 2007

      Harlan asks:

      “..from the USER’S perspective, what are the benefits of COM add-ins vs XLL add-ins? My big gripe against XLL add-in text functions is that they can only return up to 255 characters. If COM add-in text functions can return 32767 characters, that’d convince me of their value. If not, why should I prefer COM to XLL add-ins?”

      What do you think? Note the USER bit.

      Heres my view:

      Yes a COM dll can return a long string (to VBA) working with long strings is a bit of an adventure in Excel cells (and text boxes), even with VBA. XLLs in Excel 12 can also return long strings finally, (and cells can display more than 1k characters).

      As a developer I think you can pretty much get the same functionality using either approach, but generally the xll way will be more painful, more likely to crash Excel or leak memory, and run either faster or waay faster. So the USER POV is maybe COM add-ins can be cheaper, or more quickly developed, more reliable (?), but maybe run a bit slower. For a VBA user POV COM is something any VB dev understands. To use xll functions from VBA you need to either use Declares to the xll or, Excel devs won’t sweat this but pure VB folks might need some hints. UI is tons easier in COM, anyone remember XLM dialogs?

      There are also a few gaps in xll functionality (its still based on the C API from Excel 97 – so most of that new stuff isn’t available, or at least not documented). Mind you there are few gaps in the COM object model too. xlls are easy to deploy, no registry access required. The xll stuff got a big refresh in Excel 2007, and is still the recommended add-in approach. (XLLs are the add-in technology of choice for Excel)

      Summary: from a users POV direct differences are minor. From a Dev POV xlls are generally more effort, maybe this is why so few people are still doing them. In fact who, apart from City type financial institutions, and SW vendors gets xlls written? (Its a popular search term to get to codematic, but I’ve only had the odd sniff of work.)

      I’m thinking VB6 for COM btw, but I guess the same applies to .net.

      Thats my view, what do you think?



      One five oh

      Sunday, 19th August, 2007

      This is the 150th post in the 8 months since Jan – blimey. There has been just short of 1,200 comments. The most popular posts were 1 million rows about xl12 and the spreadsheet disadvantage, about specific issues with spreadsheets.

      Other blog stats – average around 200 visits per weekday,  50 or so at weekends. I don’t bother too much with this as I don’t think its a mass market topic area. The great thing for me is the quality of the comments which I think is superb, (thank you all).

      Akismet has saved us all from almost 5,000 spam comments.

      I’ve learned a ton of stuff and I hope you have all found it useful. If you have any topics you would like to get covered leave a comment. If you have never commented before, why not drop a quick hello on this post.



      Firefox blocked

      Friday, 17th August, 2007

      So soon after our browser discussion I read this on slashdot today:

      Its about a web site that deliberately blocks Firefox because some users have the ad blocker installed. Their logic being the ads pay for the content, so not viewing them is akin to stealing the content. Some great debate and insightful comments at /. as usual.

      I tried to look at the site but it just got /.’d so is unreachable at the moment even from IE.

      do you feel bad for not viewing other peoples ads?

      Also I found this site with some usage stats:

      roughly 35% ff, 35% IE6, 20% IE7, with IE6 losing share to ff and IE7. They also discuss how their stats may be biased.