https://learn.microsoft.com/en-us/azure/virtual-machines/windows-in-place-upgrade

and running PS command 

.\setup.exe /auto upgrade /dynamicupdate disable

and selecting Windows Server 2022 Datacentre as the version I wanted to install (because it is!):

I would then encounter this:

and obviously I didn't want to proceed with "nothing"!

Well, after logging a support ticket and a lot of very friendly but unhelpful email exchanges, I thought I should try again and capture screenshots of every step. And then I solved it... I had simply chosen the wrong image for my server. When I selected "Windows Server 2022 Datacentre (Desktop Experience)" I was not offered the options to keep or not keep at all - it simply went ahead with the upgrade as shown:

Two days later... actually it failed the first time - it bailed out silently, with server remaining unchanged, and nothing diagnostic in the event log that I could see. Same again second time. But the third time I ran it, the upgrade completed and my server is now Windows Server 2022 with everything seeming fine.

Most of the specific information about works of music and writing are published using the ‘blog’ tool, which lends itself to searching by categories, dates etc. In fact the site has three blogs, one for news, one for musical works, one for written works, and the last two of these are managed within folder tenants (using cloudscribe’s multi-tenancy capabilities). However by default cloudscribe uses the URL segment /blog/ in pages created in this way, and some people had remarked that this looked a bit odd having the /blog/ segment in pages about specific musical works, e.g. https://peter-tranchell.uk/writings/blog/jottings-of-a-domestic-bursar

After a little investigation, I was able to change the music and written works blogs to use the URL segment ‘works’, by changing this line in RoutingAndMvc.cs:

routes.AddBlogRoutesForSimpleContent(new cloudscribe.Core.Web.Components.SiteFolderRouteConstraint());

to this:

routes.AddBlogRoutesForSimpleContent(new cloudscribe.Core.Web.Components.SiteFolderRouteConstraint(), "works");

Having made the changes, the above post is now at https://peter-tranchell.uk/writings/works/jottings-of-a-domestic-bursar

Having done this, I needed to put in place some redirect rules so that any bookmarks or published hyperlinks that used the old URLs would automatically redirect to the new ones. This was achieved with the URL Rewrite module for IIS, and a rule like this:
<rule name="Redirect writings blog to works" stopProcessing="true">
     <match url="^writings/blog/(.*)$" />
     <action type="Redirect" url="/writings/works/{R:1}" redirectType="Temporary" />
</rule>

With all this done and seeming to work OK, I then used a link checker to see if I’d left any broken links in the site by accident. For this I used a tool that I remembered from many years ago (I found it in some old notes): https://www.deadlinkchecker.com/website-dead-link-checker.asp

This checks up to 2000 hyperlinks free of charge. It did find plenty of broken links, where external websites had moved or removed things, e.g. this History of music at Eastbourne College, which I rediscovered with some searching:

https://www.eastbourne-college.co.uk/wp-content/uploads/2020/09/musichistoryupdated-2020-09-30.pdf

I had linked to this document because it contains some information about Peter Tranchell’s brief spell at Eastbourne College in 1949-50:

Brian Polden (S48) remembers Peter Tranchell: ‘He brought with him an outrageous and infectious sense of fun and was a most gifted musician’. Ian Fraser (Sykes) (P47) remembers: ‘He was a bizarre character, particularly in his very eccentric manner of dress, including multicoloured dressing-gown sashes with tassles instead of a belt, which didn’t go too well with the fur of his academic gown. I vaguely remember the strange music he wrote for one of the school plays, which included a ‘jungle piano’, a small wooden box with pliant metal bars attached which you twanged’.

But none of the broken links had been created by my own actions – phew!

It’s a Netgear ReadyNas2, and I’m certainly not ready to consign it to the scrapheap just yet, thanks! (2.7TB of storage in a RAID configuration for safety).

The admin page is accessed at an address like https://192.168.1.xxx/admin/ (often accessed by clicking the “Admin Page” button in the “RAIDar” app, which opens the address in your default browser).

This is what I was seeing:

In Edge the detail of this feedback was as follows:

This site can’t provide a secure connection

192.168.1.111 uses an unsupported protocol.

ERR_SSL_VERSION_OR_CIPHER_MISMATCH

And on clicking the “Details” button:

Unsupported protocol

The client and server don't support a common SSL protocol version or cipher suite.

Some Googling and Binging shows a million other people having similar problems, but I didn’t readily spot a solution that worked for me… installing old versions of browsers doesn’t seem to be possible or sensible any more.

I failed to persuade Edge or Chrome to load the page, despite asking “Internet Options” to accept TLS1.0 and 1.1 (etc).

I found a solution with Firefox advanced settings. Browse to about:config

CTRL+F to look for “security.tls.version.enable-deprecated”.

Toggle this setting to “true” (e.g. by double-clicking the row).

After this, the ReadyNAS Admin page loads just fine!

I hope this helps someone, and please feel free to leave comments to improve this post!

To cut a rather long story short, I got connected by following the “Quick Setup” wizard, with the following options:

then enable VLAN ID and set it to 101.  Also note on this screen: even if it shows a remembered password, I found it was best to paste in the password again – I had some failures when I did not do this:

then create/adjust WiFi networks (mine were already set up and I didn’t need to change anything):

and we get:

The crucial thing here was enabling VLAN ID and setting it to 101.  I’ve no idea what that is and why it needs to be set to 101 - I found it out from this thread https://community.plus.net/t5/Fibre-Broadband/Replaced-Plusnet-Hub-with-Modem-amp-Router/td-p/1372911 (browsing in desperation on my phone over the terrible signal we have in Brecon!). To be fair it also says this on https://www.plus.net/help/broadband/broadband-connection-settings/ but that page wouldn’t load on my phone.

I’m thoroughly underwhelmed by Plusnet’s technical support people for not telling me this during three support calls. They stuck to the party line of trying to make me use their own router (free for £6.99 delivery and a delay of a few days waiting for it to arrive) and when I insisted I’d rather use my own they send out “instructions” for 3rd party routers which tell you only how to enter your username and password, and nothing about this VLAN ID.  Needing to set the VLAN ID is an obvious gotcha that they must be fully aware of.  And one of the emails they sent me contained information that was completely incorrect (instructions for ADSL instead of fibre).

Once this had been done, visiting the Advanced Settings showed the following:

but note – I found that changing anything on this page broke the connection completely, even returning the settings to these original values. I then had to go back around the Quick Setup routine, re-entering the password and re-setting the VLAN ID.

This is not the place to talk about how to use LetsEncrypt, but if you have a web server with only one external i.p. running multiple sites you have to certificate them all at the same time, and suffice it to say that one site on our server was proving stubborn... my Orchard CMS site https://www.patranchell.info/ or rather, as it was then, http://www.patranchell.info/

LetsEncrypt's validation process includes creating some special files in a known location within the website and checking they can be accessed in that location from the internet, but this was failing because by default Orchard CMS will not serve up static files, and certainly not extensionless static files. The solution turned out to be pretty simple changes to my web.config (after some trial and error!) shown as insertions below:

        <handlers accessPolicy="Script,Read">
            <!-- Clear all handlers, prevents executing code file extensions or returning any file contents. -->
            <clear /> 
   <add name="LetsEncrypt" path=".well-known/acme-challenge/*" verb="*" modules="StaticFileModule" preCondition="integratedMode" resourceType="File" requireAccess="Read" />
Once this had taken effect, the LetsEncrypt tool was able to validate and convert my site, automatically adding the https binding. But my site still wouldn't load under https. Hhmpphh.

I then had to install a new Orchard module (I'm running v1.9.2 btw) called "Secure Sockets Layer". And then I had to visit the settings for this module:

and set the following (note that most of these settings only appears after ticking the first box):

The result...free https encryption, which should auto-renew for ever! Amber Rudd will be upset!

In two hours I had a domain registered and the site established (using mojoPortal CMS) at http://brecon-maths-physics-tutor.org/  running on hosting from exeGesIS SDM.

It's super basic, but gives the essential info, with social sharing, and an enquiry form. I need to do some work on the cosmetics, but that is very time-consuming and I don't have time at the moment.

Now to get it visible on the internet!

First we added a QR code to the poster so people could easily open the site on their phones.

But what about the search engines? The problem with common services like tutors, dentists, driving schools etc is that the search engines seem to be flooded with the national chains and agencies, plus pay-per-click ads for these. I'm not sure how, but they spoof local relevance so that when you search for "brecon car hire" they make their results say "Car hire in Brecon - best prices" even though they may be a national chain without a branch in Brecon.

So with more optimism that expectation I made the site content and metadata as sensible as I could with the limited material, and set up the site in Google and Bing Webmaster Tools.

To my amazement, the following morning the site was on page 1 of Bing when searching for "Brecon maths tutor". Nowhere to be seen in Google. Wow for Bing! Or so I thought.

A week later, after changing absolutely nothing, the site has vanished without trace from Bing, for reasons I cannot fathom, and is appearing high up page 2 in Google. Interesting, watch this space!

Figure 1: a sheep

It's good, but not as good as OpenLiveWriter. For example, image handling isn't as easy (e.g. we cannot add alt text or title on the img element) and I've not found a way to edit the HTML source directly.

Both editors can do tables, and while the handling is rather different they have similar limitations – neither editor seems to be able to create <th> elements (headings):

The range of "Insert" tools is rather different, each having strengths for some purposes. Here's OLW:

And here's Word:

While the range of tools is greater, they are less helpful for blogging, I'm also not convinced they will all work when posting through the MetaWebLog API.

To test this, let's try some SmartArt (a labelled pyramid):

And a chart:

Some symbols: ₯©℅ﷺ

A lightning bolt shape:

And some WordArt (this should appear blue with reflections below the letters):

Now let's "publish" and see what happens!

Overall, I'd probably use Word if OLW didn't exist, but it's capabilities are less web-friendly.

This is my first post – configuration was the same as with the old WLW, and I’ve not spotted any differences yet apart from the shortcut icon and the name. Oh, and the complete absence of all those nice plugins, but the website says they are coming soon.

I re-did all the panel connections, but we still ended up with one bad line with interior-grade cabling traversing the exterior of the house just high/far enough that I didn’t want to attempt replace it, and it seems to cause drop-outs for all devices on the network).

Installation of the Devolo could not have been simpler, taking about three minutes and working first time without issue; so far the connection has been completely stable.

But what about speed?  I did a quick test, copying a folder containing about 1Gigabyte of files from a local drive on the PC to a NAS drive across the network. Here’s the contents of the folder:

Here’s the result over the Devolo “1200 Mbps” connection (which translates in theory to 143MB/s!):

Average speed over the Devolo Powerline 1200+ around 8MB/s (or 67 Mbps).

And here’s the results over the (unreliable) ethernet connection:

Average speed around 11MB/s (or 92Mbps), that’s 37.5% faster.

I then tried further tests using Totusoft’s free “LAN Speed Test (Lite)” from http://totusoft.com/downloads/

Here’s the Devolo:

and here’s the Ethernet:

The results are similarly in favour of the ethernet connection. 32% faster write speed, and 14% faster read speed.

I’m a bit disappointed.

However, I did get this result a number of times with the ethernet connection – “Error reading test file” - illustrating how unreliable it has been:

Our greatest concern is stability rather than raw speed, as family members have all been going nuts with the drop-outs – I had managed to prove that whenever the line to the study failed, other devices in the house would suddenly lose their WiFi and/or ethernet and/or internet connection, for reasons I don’t fully understand. So let’s see over time whether the Powerline solution turns about to be more stable. After two hours the answer is seeming to be “yes”!

The topic is… how to manage all the photos we take with camera and phone, and how to share these with family and friends (but not the whole world).

Facebook or Instagram, Flickr or Picasa, surely!? Umm, no actually we use the free version of Gallery Server (“GS”) to share photos and videos with family and friends (requiring a log in) and to share certain categories of photos with anyone (e.g. from the Pen y Fan Race).

This post is not the place to discuss the reasons why, but essentially it’s about retaining control and minimising overall costs. We run Gallery Server on cheap but decent hosting from Arvixe, using their personal class ASP.Net hosting package. This offers unlimited storage space and data transfer, plus “full trust” ASP.Net hosting, which is good for GS.

It’s important to state that we are using GS and Arvixe for publishing selected photos, and not as the primary repository of the originals – that really determines the workflow here. If we were simply pushing every photo onto Arvixe/GS and publishing the lot then things would be much simpler, but the bandwidth requirements for the upload and the time taken would be more than our current internet connection can bear (no fibre yet).

So our primary repository is on a NAS drive on the home ethernet network, specifically a Netgear ReadyNAS Duo v2 with a couple of TB drives in RAID5 configuration (for resilience). These files are in turn backed up every night to an older Netgear ReadyNAS Duo (v1) that used to have the same disk arrangement until one went pop, so no RAID any more!

So first we need to get the photos into the primary repository.

Cameras with memory cards

Plug the SD card into the main PC and copy the files into the relevant year/month folder on the NAS drive, e.g. P:\FlowerPhotos\2015\08-August  (we always create a new folder for each month in the format nn-monthname which is readable and sorts correctly).

Phone

Let the phone sync the photos to OneDrive, which then syncs them automatically to the main PC > copy the files to the folders mentioned above.

Cataloguing

We use ThumbsPlus v10 to catalogue and view photos within the home network. Although this app has gone through some turbulent times since the upgrade from the excellent version 7, it’s still a useful tool with some strong points, in particular the ability to add custom attributes for files and batch process selections with custom processing steps. We make use of both of those facilities in this workflow.

In ThumbsPlus we scan the folder(s) where the new files have been added to make thumbnails (catalogue entries). Then we whizz through them “tagging” the ones we want to make visible to others (this is a single keystroke while viewing each image, so very simple). We might also delete the no-hopers at this stage. When we’ve reviewed the folder, we select all the tagged thumbnails in the folder (Edit > Select by > Tagged) and add a more permanent attribute, Ctrl+U to “assign user fields” for the selected images, then enter “Y” into the “ForWeb” attribute (this adds the value to all the selected files at once). Repeat for any other folders involved.

Then we pull together all the files that have been flagged as ForWeb=Y with a “Found files” query (one of ThumbsPlus’ many items of bizarre nomenclature), designed as shown and stored/named as “FlowerPhotosForWeb”:

To refresh the list in this query we just right-click on it and “Requery Now”.

Then we select all files (Ctrl+A), then we run a batch process to resize the images to not more 1024x1024 and drop them out into a new folder structure mirroring the original, but ready for uploading to the web server. The steps in the batch process are as follows:

and the first batch step in detail (choosing the “Downsample” interpolation is really important as the others give poor results):

and the second batch step:

and the output options (including choosing 90% JPEG quality):

This takes a few seconds/ minutes to run depending on how many new images have been added (it doesn’t re-process any that are already in the output folder unless the “Overwrite files with the same name” option is ticked on the last screen above. We find that the original folder structure with the year/month folders is preserved below the output path above, but containing only the selected images at the appropriate size.

The manual work has now finished, and the rest will happen all by itself, though it’s possible to push it along quicker in various ways if needed.

Moving the files to the web server is achieved with SyncBack Free, which, as the makers say, is the “best free Windows backup program on the planet”. It’s necessary to create a “profile”, and make lots of decisions for this profile, like where it should work from and to, when it should run, etc. In my case, it’s doing a simply backup from my P:\FlowerPhotosForWeb\FlowerPhotos to the FTP address of my hosting account, and putting the files into the “gs\mediafiles\family photos” subfolder below the GS web site folder. The “backup” setting ensures that any files that are in the source but not the destination, or any files that have changed at the source, are copied to the destination. Any files that are in the destination but not the source are left in place (which might happen if we add files to GS by other means). The finished article looks like this:

With SyncBack we can either let a profile run on a schedule, or run it interactively – I have this scheduled to run every evening. I should say at this point that a year or so ago I had Arvixe admins emailing me saying I wasn’t allowed to use their hosting as a backup service, as they had detected automated FTP file transfers and too much storage use within my account. I replied that I was not using their service for backup purposes, but simply to populate family photo albums efficiently, and asked them to clarify what aspect of “unlimited storage” I was infringing, and they went silent. Material for another post here methinks.

So now the files are on the web server, there’s one more step to make them appear to friends and family… GSP needs to “synchronise” in order to recognize and process the new files. As with SyncBack, this can be done manually (Actions > Synchronize), or automatically (Site admin > Albums > General > Synchronization). Ours is set to synchronize once per day.

Because we’ve resized the images before uploading them to GS, we can ensure that no further processing/re-sampling is done by tuning the settings. GS will always create thumbnails of chosen size for each media object, but will also create optimised/reduced versions of each file if settings dictate, either keeping or discarding the original as you wish. In our case we don’t want it to create a new optimised version, unless we use some other workflow to load out-of-camera large files. The resized image files generated above are normally 2-500KB, and not more than 1024 pixels in either dimension, so we set the following in Site admin > Media objects > Images:

Once this synchronization has run, any new photos appear in the gallery. This can mean that an image takes a couple of days to get from camera or phone to the website, and there is a manual filtering process needed that takes a minute or two for a folder of images. If we want to speed things up, the SyncBack and GS process can both be initiated manually instead of waiting for them to run automatically, so overall this workflow seems OK, but we are always looking to improve, and the clunky bit to me is currently the ThumbsPlus part.

After a bit of binging and even some googling I ran sfc /scannow from an administrative command prompt and got this:

and the moment I re-started outlook my emails were sent, yay.

The fact that others were reporting this during testing of early beta releases of W10 makes me think they could have built this fix into upgrade, but there we go. Hope this helps someone.

Setting up

1. A turntable connected to an amplifier. In my case, an old (but still lovely) DUAL CS 505-3
   
   with a Cambridge Audio Azur 551P pre-amp, and a Cambridge Audio Topaz AM1 amp.
2. A connection from the turntable to the computer. I have a Xitel gizmo (currently called the “Xitel INport Deluxe Analog to Digital Audio Recording Kit”) that converts from normal red/white HiFi leads to a USB port on the PC.
   
   NB don’t use the recording software that may come with this device – Audacity is much better!
   It would be even easier with a USB turntable, but I'm told quality suffers.
3. A PC running Audacity. And plenty disk space.

There might be some footering about to ensure that you have the correct recording input in Audacity. Mine has to be set to “Line (INport)”.

Set the recording level so that there is no clipping. Do this by starting recording, and adjusting the Input Volume. I have it at 0.52 at the moment, though it could probably safely be a bit higher. The important thing is not to have it too high, as that leads to distortion.

Recording

Once the levels are OK we are ready to record. Start recording, then drop the needle, and sit back to enjoy the music. Here we are recording Melanie’s Garden in the City album (complete with scratch ’n’ sniff “smell Melanie’s Magic Garden” sleeve!):

At the end of the first side, hit Pause (the top left button in the screenshot above). After turning over, but before dropping the needle, I hit Pause again to restart recording, then drop the needle.

And the end of the LP, hit Stop (third button along).

Cleaning up

Now it’s time to clean up the track before exporting. At present it is not saved to disk in any place you will find it again. It you don’t have time to complete the clean up and export job before shutting down, then you should save the Audacity project (File > Save Project…). If doing everything in one go, I usually don’t bother saving at all, not least because after saving the project to disk everything becomes much slower, as it seems to write all changes back to disk automatically (at least, the disk access light on my NAS drive flashes when I do anything).

The first clean-up task I run is click removal, because almost all old LPs have a few big clicks. This doesn’t remove every one of them but I think it helps, without collateral damage.

Effect menu > Click Removal… (I accept the default settings):

At the beginning of each side of the LP, I remove the initial clunk of the needle drop by selecting and deleting that section (click and drag over the track, then hit the Delete key):

Now we can clean up the recording a bit more by removing the stuff you can see in the screenshot above before the music starts – in my case some hissing and crackling and a low throbbing hum. To do this, select a couple of seconds’-worth of the noise:

Then go to Effect menu > Noise Removal… and click “Get Noise Profile” button:

The dialogue closes, leaving you slightly unsure whether anything has happened, but don’t be perturbed. Now select the entire recording (Ctrl+A), and open the Noise Removal dialogue again. This time, click “OK”

When it has finished, the grunge before the music should have vanished:

Next we need to identify and mark out the tracks. This is done by clicking into the track a short time before the track starts, e.g. as shown below, and using Ctrl+B. Things should look this as a new label appears:

Simply type the track name. You can click back into this later to edit, though it is a little cantankerous (e.g. sometimes it refuses to accept the letter “k”!).

Now move through the recording identifying and labelling the beginning of each track. You should be able to play back sections simply by selecting them and hitting Play.

Exporting

When all tracks have been labelled and any clunks and gaps removed, it’s nearly time to export. But first, create some metadata for the album as a whole.

File menu > Edit Metadata…

edit the metadata to set the album title and artist:

Now we are ready to export the tracks to disk in their final form:

File menu > Export Multiple…

and change the destination folder and other options if needed. Here I am exporting as MP3 into the magic folder that is picked up automatically by iTunes:

You will then be taken through the metadata for each track where you can ensure the track name and number are correct:

Then the export begins:

And that’s it. You now have your album/single saved out as files ready to play in iTunes or on the SONOS, or whatever.

These days everything that can be opened is made to close itself again automatically. Doors have heavy weighted thingies so they can smack you in the mouth, apps close themselves when they think you have finished using them, and Microsoft SQL Server databases have a setting that makes them close down when all users have disconnected.

One of the web sites I run (www.mynydd-du.org.uk) has been performing terribly for the last two weeks – frequent alerts that the site was down, and terrible page load times. The site is built with mojoPortal CMS, using a SQL Server database. After a lot of frustrating exchanges with the hosting company (Arvixe), who could see nothing wrong, I finally spotted that this “Auto Close” setting was set to True for this site’s database, and False for all my others. My account didn’t have permissions to change this setting so I’ve no idea how it got like this, but that’s how it was. I explained this to Arvixe, and waited… eventually they made the change, after I went to their main quality assurance people - thanks to Joe Audette at mojoPortal for giving me the crucial email address.

Here’s the difference it made, as measured by Uptime Robot:

The setting was changed at about 03:40, and apart from one bad reading the average response immediately time changed from about 8 seconds to about 0.5. And in the time since, I’ve only had one alert of the site being down, instead of LOTS in every preceding period.

So the moral of the story is… in SQL Server databases always set Auto Close to False, and in general, don’t trust anything that closes itself automatically.

*** UPDATED SOLUTION - MAY 2015 ***

As noted in one of the comments below, there is now an single Canon installer that should resolve this.

Visit this page and select your operating system from the dropdown list.

A list of downloadable installers should appear, including

MP620 series Mini Master Setup

md6l-win-mp620-1_05-ea24.exe (file name will be slightly different for other OS).

Download and use this installer - it worked a treat for me on a new Surface Pro 3, which could not see my networked MP620 at all until I ran this, and now prints and scans perfectly.

*** END UPDATE ***

After some investigation I found some mentions of this on the web, but no solutions. My first thought was to update the printer drivers. After some hunting around I found version 1.04 for 64-bit systems on Canon Canada’s site (downloading file: md64-win-mp620-1_04-ea24.exe). But after running through this installer (a self-extracting zip that then launches its setup.exe) following likely options I found that the normal Windows 7 “add printer” mechanism could not discover my printer on the network at all. I could browse to it in Windows Explorer, and on first access it installed itself but with Microsoft drivers dating from 2006. And it refused to allow this driver to be updated by browsing into the extracted folders from the downloaded file.

Getting a bit frustrated, I then spotted some small-print on the Canon download page. In the “Caution” section, it says:

2. The users who use the captioned models through a LAN connection cannot perform scanning after upgrading the MP Drivers, or uninstalling and reinstalling them. Preventive action: To reinstall those MP Drivers while already using a LAN connection,manually specify FILE as a port in the Installation Complete window.Then, download the latest version of Network tool to reperformthe network setup [sic].

So I cleared the new drivers off again using the “Canon MP620 series” > “MP Drivers Uninstaller” command under my Start menu. Then I ran the downloaded driver installer again, and sure enough the final screen allows you to tick “Select printer port” and click “Manual Selection”…

… then pick FILE:

… and then Complete.

I was then a bit stumped as to what it meant by the “Network tool”, but a bit more reading around suggested this was the “Canon IJ Network Tool”. This was found on this Canon Asia page: http://support-sg.canon-asia.com/contents/SG/EN/0200026301.html with file name ldtl-win-mx850-2_5_7-en.exe

This too was a self-extracting zip which launched a wizard:

it detected the printer on the “FILE” port (ignore the first row in this shot – this is because I went round the process twice to get these screenshots):

then…

and that was it – scanning and printing now working perfectly over the LAN. I hope this helps someone!

It was failing with error code 0x81000039. Some searching led me up a few garden paths, and then to this post:  http://www.wincert.net/tips/microsoft-windows/2072-windows-backup-fails-with-error-code-0x81000039.html

I tried it without much hope, because the post suggested this was to do with accessing a drive on a Windows XP machine…

Then I restarted my backup job… it worked!  I don’t think I’d come across Wincert before, but they seem to know what they are talking about, and there are some good tech posts on their blog.

My favourite web authoring tool, however…

… since an automatic update to Windows Live Essentials the other day, Windows Live Writer has failed to start, giving “Windows Live Writer has stopped working” immediately. Actually I’m not 100% it was working just before that, but I think so.

I am of course writing this in Windows Live Writer, so what was the solution?  I tried Microsoft “Fix it”, which didn’t. Endless googling/binging had yielded nothing helpful, and the event log gave no clues. However, I also found that Adobe Reader’s update to version “X” would not install because of “Error 1327. Invalid Drive F:\”.

I found an article on the Adobe forums suggesting re-creating the missing drive using (for example)

subst f: c:\windows

and the installation worked fine. And this also fixed Windows Live Writer!

Obviously I don’t want to have to do that every time, so I’ve mapped F: to a folder on the NAS. And in fact this is how it was many months ago, until at some point I killed my F drive in favour of an S. This must be behind the problems – something in the guts of WLW (and the Adobe registry settings too?) must be referencing the ghost of my old F drive, and the recent updates must have changed the behaviour to make this critical.

Anyway, problem solved, though I am left feeling uneasy that I don’t know why this happened.

Some time later… I couldn’t let this lie… I found a couple of registry keys set to “F:”

HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Explorer\User Shell Folders\Personal

and

HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Explorer\Shell Folders\Personal

I changed these to “%USERPROFILE%\Documents” and “C:\Users\Flowers\Documents” respectively, then disconnected my F: drive again, and sure enough WLW opens fine. So that is probably the true fix to this problem: WLW crashes on start-up if either of these keys points to a bad drive.

Now to show Helen how to use WLW to post to her blog…

After entering about 3 books it became apparent that this was a pretty boring activity, and my idea of getting the kids to do it was not well received even by Lois. So I hunted around for a web service that could do it for me with minimal effort. I found isbndb.com and all my troubles were over. I’ve added a magic button to my data entry form to fetch data from the ISDBdb web service - all you have to enter is the 10-digit ISBN code, and the other fields are magically filled in for you.

It’s quite a powerful API, allowing searching in a variety of ways and returning data about books, authors, subjects etc. For the moment I’m only using a simple ISBN look-up in an XMLHTTP request, where the request looks like this:

returning XML data like this:

<ISBNdb server_time="2012-02-12T19:58:11Z">
  <BookList total_results="1" page_size="10" page_number="1" shown_results="1">
    <BookData book_id="archaeology_a33" isbn="0500276056" isbn13="9780500276051">
      <Title>Archaeology</Title>
      <TitleLong>Archaeology: theories, methods, and practice</TitleLong>
      <AuthorsText>Colin Renfrew, Paul Bahn</AuthorsText>
      <PublisherText publisher_id="thames_and_hudson">New York : Thames and Hudson, c1991.</PublisherText>
      <Details change_time="2005-03-31T08:09:28Z" price_time="2011-12-31T01:26:57Z" edition_info="" language="eng" physical_description_text="543 p. : ill. (some col.)" lcc_number="CC165" dewey_decimal_normalized="" dewey_decimal="" />
      <Summary></Summary>
      <Notes>Includes bibliographical references (p. 493-524) and index.</Notes>
      <UrlsText></UrlsText>
      <AwardsText></AwardsText>
      <Subjects>
      <Subject subject_id="archaeology">Archaeology</Subject>
      </Subjects>
      <Authors>
        <Person person_id="renfrew_colin">Renfrew, Colin</Person>
        <Person person_id="bahn_paul_g">Bahn, Paul G.</Person>
      </Authors>
    </BookData>
  </BookList>
</ISBNdb>

So it was a relatively simple matter to parse this XML (using MSXML) and bung the values into the right places. Actually it is not completely simple, because the data from the web service is rather variable, and doesn’t correspond directly with my data structure; for example, you can see from the example above that the PublisherText node contains a string that I would consider the publisher, publishing location, and publishing date, all mashed up together – slightly disappointing from a supposedly structured database. But hey it’s pretty good, and the result is we can now enter a new book in a few seconds.

My Windows 7 64-bit PC failed to install an update to Paint.Net a few weeks ago – mildly irritating at the time, but in fact just the first in a long line of problems with the same underlying cause. I then started noticing that TrustedInstaller.exe was running at around 50% of my CPU for hours at a time (one core of two), and my Event Log (Event Viewer > Windows Logs > Application) was getting loads of “Information” messages from “Windows Error Reporting” like the following:

Fault bucket , type 0
Event Name: WindowsWcpOtherFailure3
Response: Not available
Cab Id: 0

Problem signature:
P1: 6.1.7601
P2: base\lstring\lblob.cpp
P3: RtlTranscodeLBlobs
P4: 2221
P5: c0000161
P6: 0xe2a94a52
P7:
P8:
P9:
P10:

Attached files:
C:\Windows\Logs\CBS\CbsPersist_20111127072435.cab
C:\Windows\Logs\CBS\CbsPersist_20111202195623.cab
C:\Windows\Logs\CBS\CbsPersist_20111206223453.cab
C:\Windows\Logs\CBS\CbsPersist_20111210112002.cab
C:\Windows\Logs\CBS\CbsPersist_20111213214632.cab
C:\Windows\Logs\CBS\CBS.log
C:\Windows\servicing\Sessions\Sessions.xml
C:\Windows\winsxs\poqexec.log
C:\Windows\System32\LogFiles\Scm\SCM.EVM
C:\Windows\Logs\CBS\FilterList.log

These files may be available here:
Analysis symbol:
Rechecking for solution: 0
Report Id: f8c2fdf4-28a3-11e1-8771-001a4d536a18
Report Status: 0

Also, when I tried upgrading Office 2007 to 2010, it just bailed out and rolled back every time.

After scouring the web for answers, I found lots of people having similar problem with Vista and Windows 7, with many saying this would require a re-installation of the operating system. But also I quickly found the “System Update Readiness Tool for Windows 7 for x64-based Systems” and downloaded from here http://www.microsoft.com/download/en/details.aspx?id=20858

Running this outputs this file C:\Windows\Logs\CBS\CheckSUR.log which records what problems were found and whether they could be fixed.

I was getting these lines:

…

Checking Component Store
(f)    CSI Corrupt Identity    0xC0000161    (not UTF8)    amd67_microsoft-windows-i..stomloggi/gbinaries_31bf3856ad364e35_6.1.7600.16385_none_0c741d2872a20a44    Bad identity
(f)    CSI Manifest Missing    0x00000002    amd67_microsoft-windows-i..stomloggi/gbinaries_31bf3856ad364e35_6.1.7600.16385_none_0c741d2872a20a44.manifest    amd67_microsoft-windows-i..stomloggi/gbinaries_31bf3856ad364e35_6.1.7600.16385_none_0c741d2872a20a44   
(f)    CSI Missing Identity    0x00000000    identity    amd64_microsoft-windows-i..henticationbinaries_31bf3856ad364&35_6.1.7600.16385_none_39dd2292"22c1d9e   
(f)    CSI Could Not Open file    0x0000007B    winsxs\Manifests\amd64_microsoft-windows-i..henticationbinaries_31bf3856ad364&35_6.1.7600.16385_none_39dd2292"22c1d9e.manifest    amd64_microsoft-windows-i..henticationbinaries_31bf3856ad364&35_6.1.7600.16385_none_39dd2292"22c1d9e   
(w)    CSI Replacement File: CreateFile Failed    0x0000007B    winsxs\Manifests\amd64_microsoft-windows-i..henticationbinaries_31bf3856ad364&35_6.1.7600.16385_none_39dd2292"22c1d9e.manifest    amd64_microsoft-windows-i..henticationbinaries_31bf3856ad364&35_6.1.7600.16385_none_39dd2292"22c1d9e   
(w)    CSI Replacement File: CreateFile Failed    0x0000007B    winsxs\Manifests\amd64_microsoft-windows-i..henticationbinaries_31bf3856ad364&35_6.1.7600.16385_none_39dd2292"22c1d9e.manifest    amd64_microsoft-windows-i..henticationbinaries_31bf3856ad364&35_6.1.7600.16385_none_39dd2292"22c1d9e   
(w)    CSI Replacement File: CreateFile Failed    0x0000007B    winsxs\Manifests\amd64_microsoft-windows-i..henticationbinaries_31bf3856ad364&35_6.1.7600.16385_none_39dd2292"22c1d9e.manifest    amd64_microsoft-windows-i..henticationbinaries_31bf3856ad364&35_6.1.7600.16385_none_39dd2292"22c1d9e   
(w)    CSI Replacement File: CreateFile Failed    0x0000007B    winsxs\Manifests\amd64_microsoft-windows-i..henticationbinaries_31bf3856ad364&35_6.1.7600.16385_none_39dd2292"22c1d9e.manifest    amd64_microsoft-windows-i..henticationbinaries_31bf3856ad364&35_6.1.7600.16385_none_39dd2292"22c1d9e   
(w)    CSI Replacement File: CreateFile Failed    0x0000007B    winsxs\Manifests\amd64_microsoft-windows-i..henticationbinaries_31bf3856ad364&35_6.1.7600.16385_none_39dd2292"22c1d9e.manifest    amd64_microsoft-windows-i..henticationbinaries_31bf3856ad364&35_6.1.7600.16385_none_39dd2292"22c1d9e   
(w)    CSI Replacement File: CreateFile Failed    0x0000007B    winsxs\Manifests\amd64_microsoft-windows-i..henticationbinaries_31bf3856ad364&35_6.1.7600.16385_none_39dd2292"22c1d9e.manifest    amd64_microsoft-windows-i..henticationbinaries_31bf3856ad364&35_6.1.7600.16385_none_39dd2292"22c1d9e   

Summary:
Seconds executed: 688
Found 4 errors
  CSI Could Not Open file Total count: 1
  CSI Manifest Missing Total count: 1
  CSI Missing Identity Total count: 1
  CSI Corrupt Identity Total count: 1
  CSI Replacement File: CreateFile Failed Total count: 6

Unavailable repair files:
    winsxs\manifests\amd67_microsoft-windows-i..stomloggi/gbinaries_31bf3856ad364e35_6.1.7600.16385_none_0c741d2872a20a44.manifest
    winsxs\manifests\amd64_microsoft-windows-i..henticationbinaries_31bf3856ad364&35_6.1.7600.16385_none_39dd2292"22c1d9e.manifest

Now this all looked rather suspicious, and after checking files on a Windows 64 bit laptop, I concluded that “amd67_microsoft-windows-i..stomloggi/gbinaries_31bf3856ad364e35_6.1.7600.16385_none_0c741d2872a20a44” was surely supposed to be

“amd64_microsoft-windows-i..stomloggingbinaries_31bf3856ad364e35_6.1.7600.16385_none_0c741d2872a20a44”  (I’ve marked the changed characters in bold and red).

So where were the bad strings coming from?  A registry search revealed this key:

[HKEY_LOCAL_MACHINE\COMPONENTS\DerivedData\Components\amd67_microsoft-windows-i..stomloggi/gbinaries_31bf3856ad364e35_6.1.7600.16385_none_0c741d2872a20a44]

Since all the despairing comments on web forums were suggesting I was heading towards a reinstallation of Windows, I decided nothing to lose and exported then deleted this registry key.

I then re-installed the “System Update Readiness Tool” and found that the lines about the corrupt identity had disappeared.

After a reboot, lo-and-behold I found I could install Windows Updates without TrustedInstaller running for hours, and without the bad entries in the Event Log. Wven better, Paint.Net installed without incident. And best of all, Microsoft Office 2010 upgrade run sweetly.

Phew.  I’ve no idea what could have caused the registry corruption in the first place, or whether my PC is really fixed, but it is a lot better than it was this morning!