Beware Apple's Epson software update 2.14 (April 2013)

I just did a routine batch of software updates for my 27" Mac, including the April 2013 Epson update 2.14. This is labeled as a printer update, but it's a printer and scanner update.

After the install my Mac locked up on boot, just before it should show the mouse icon. I pulled the firewire 800 cable connected to my Epson Professional V700 scanner and the boot completed.

After some experiments showed that I couldn't boot as long as the scanner was connected I power cycled the scanner. It was behaving oddly, I had to pull the power cord. After that I was able to boot.

I'm not sure the problem is fixed, Apple's Discussion board has a number of complaints:

• epson printer software update 2.14...: Apple Support Communities
• Bad print to new Epson after OS X update: Apple Support Communities

I'll update this post if the problem recurs, but for now the trick was to pull the firewire cable on restart and then try power cycling the scanner.

See also:

• Gordon's Tech: Obnoxious old Epson Scan bug: EPSON Scan cannot be started. 2/2008. Hmm. Strangely familiar.

Update 10/21/2013

• After four months of intermittent but increasingly odd boot behaviors followed by a few days of high frequency kernel panics the flat 400/800 adapter firewire cable connecting this scanner to my firewire chain split open. So, in retrospect, this may have been the start of the firewire cable failing. In future I'm going to do a routine restart before I do updates -- just to make sure things are working before I change software. There's a lesson here about the problems of complex configurations given the limited diagnostic capabilities of OS X.

Indexing Freemind documents for Windows Search: just add extension as free text

There are a lot of software domains that disappoint me. It's a 'get off my lawn' kind of thing. Enormous excitement about cosmetic changes to iOS 7, but nobody cares about applications for knowledge visualization. Sob.

Sure, there are some good apps for OS X [1] (though even they lack innovation [2]), but my work day is spent in the software desert of Windows 7, a forgotten platform served only by minor vendors slowly degrading once half-decent products.

On the Windows 'mind mapping' front MindManager is expensive, increasingly slow and burdened by feature cruft, and locks data up in a proprietary format. I liked XMIND for a while, but then a software update slowed it to a crawl. It also seems to have been abandoned.

So, lately, I've turned to FreeMind. Not because it's open source and free, not because there's ongoing development, and definitely not because it's a Java app or "cross-platform". I've turned to Freemind because, despite being homely [3], slow to launch, and having an eclectic UI, it has users, performs quickly when it's running, and, above all, it has the closest thing to a standard file format in the industry. A file format that's plaintext.

More and more, I love plain text.

I think I can live with FreeMind -- but only if I can retrieve documents using Windows Search. I live and die by free text search. I took a look at the FreeMind IFilter ($20) for Windows Search, but I was unimpressed with the klunky install and configuration requirements. It had a bad smell.

Fortunately, there's a simple workaround. FreeMind .mm files are plain text (did I mention I love that?). All I had to do was tell Windows Search to index .mm files as free text:

That worked.

• Mac Mind mapping software: Inspiration, OmniOutliner and MindNode 7/2011
• XMind: Impressions and comments on the mind mapping market 11/2012
• XMind: Software made in China for OS X and Windows 11/2010
• Mac Mind mapping software: Inspiration, OmniOutliner and MindNode 6/2011. I used Inspiration in the late 90s, early 00s. I'd use it now but the file lock is too strong.
• Mindjet (MindManager): operational problems -> maybe improving 7/2007
• ConceptDraw can import MindManager files 8/2007. More recently I tested it, but it didn't add anything new or interesting.
• GrandView and idea management software - Fallows and more 1/2010. Most MindMapping software is an outliner with a spatial layout.
• Tinderbox, Simplenote, MindNode and data freedom 10/2011

[1] I'd love to see OMNI Group enter this market, possibly building on OmniOutliner. I use MindNode on OS X.
[2] Rereading my 10/2011 idea for implementing a graph app atop simplenote nodes I'm sad nobody has done this. If I lose my job maybe I will.
[3] Worse than homely, FreeMind currently has a major usability problem. There's nothing in the UI to tell you that a node has collapsed children. It's quite weird.

Post Google Reader: Feedbin, Newsblur, Feedly - all disappointing at the moment

I'm scrambling as usual, so no time for a full post. I've been running through Feedbin, Newsblur and Feedly today. I paid for both Feedbin and Newsblur, and Feedly is free.

My biggest disappointment is Feedbin and Reeder.app -- the sync isn't working! Feedbin says I have 270 unread, Reeder.app says 1086. Reset didn't help. Next disappointment -- both Feedbin and Newsblur didn't get my Google Reader custom names for feeds. Only Feedly picked those up.

Newsblur (open source!) is far too social. It insists on showing me comments as I try to read my feeds. It shares to Pinboard in the web app, but not in the iOS app. No Reeder.app support. Newsblur is strict folder hierarchy - no tags, no acyclic graph organization. No URL sharing. Newsblur has best performance and most features, but so far it's not right for me. I'll try again in a month or so.

Feedly has the very irritating plugin model and no Reeder.app support. It's my emergency fall back.

Feedbin has URL type share to pinboard; that's enough for me since I use Pinboard tags to control IFTTT routing to app.net, twitter, and kateva.org/sh. Feedbin's tag model is a much better fit to my GR org than Newsblur's hierarchy. That's really big for me. I'm very glad I can now 'hide' all tagged feeds. Renaming feeds are unsubscribing is very awkward; I'd argue there's a bug with that UI. (Tip: rename feeds you don't want to 'z', then when all done with name repair save, then select all 'z' then remove.)

Ugh, this will be a hard transition. There's still no true heir to GR's basic functions, not even counting all the abandoned features it had.

If not for the Reeder sync problems I'd go with Feedbin, but that's a bad fail. For the next few days I'm happily back to Google Reader (7 days left!), I'll see if Feedbin's sync bug clears itself.

Unexpected "Welcome to Mail" on OS X Lion: restoring lost email through two pref files

Emily opened her email 3 days ago and saw "Welcome to Mail".

That's bad. It meant OS X had lost track of her mail archives. The files were all on disk, but the OS knew nothing about them.

I didn't try entering new information of course. That would have been quite disastrous. Instead I poked around and found the usual advice to restore ~/Library/Preferences/com.apple.mail.plist from Time Machine. [1]

That did nothing.

Then I found Power loss, Mail accounts gone,...: Apple Support Communities, a post that recommended also restoring ~/Library/Mail/V2/MailData/Accounts.plist. That looked good, since her current file was 1kb and the old one was 6kb. After a Time Machine restore her email was back.

Of course this doesn't explain why her data was lost in the first place. HFS+ is supposed to be journaled, we should't lose these files even in a nasty crash. So now I have to run some drive diagnostics.

Update: [1]. Oops, i should have read more closely damnit. I think the SuperUser article had the right advice pre-Lion.

/System/Library/PrivateFrameworks/ScreenReader.framework/Versions/A/Resources/English.lproj/ExceptionsDatabases/com.apple.mail.plist
~/Library/Mail/MailAccounts.plist
~/Library/Preferences/com.apple.mail.plist

Personally I wouldn't mess with /System unless the other two fail and the directory changed post Lion.

CloudPull - backup your Google Cloud properties -- including your old shares and your multiple Drive accounts.

I've been using Golden Hill Software's CloudPull Premium ($15 now, I only paid 410 due to a promo, TidBITS users can jump get a 20% discount if they insist) for a month or so. It backs up parts of the Google data associated with 3 of my Google identities -- Personal, Public (jgordon@kateva.org) and Work. Sooner or later I'll start backing up the kids stuff too.

The backup data includes my Google Calendars (primary and subscription/shared as .ics), Google Contacts (3,496 as .vcf), Drive files (Google versions converted to iWork, else native) and Reader subscriptions, starred, shared and liked (not sure of format, seems a mix of database  and references). The repositories are searchable through CloudPull and, since I changed the store setting to be in my Documents folder, via Spotlight. I don't backup my email accounts; those are in Mail.app.

The app seems to work very well. It's designed to backup every 1 to 24 hours, but I just run it every few weeks. It would be nice to have a once weekly or once monthly backup option, or to be able to schedule backups for the wee hours of the morning. I don't backup 

My archive includes about 40,000 Google Reader shared articles, a SQLite database of about 400MB and total storage of about 5 GB. Not trivial, but manageable. The storage architecture makes use of SQLite and many shortcut links.

I am quite fond of this app. It has a minor glitch in an early version, the developer fixed that quickly. Highly recommended for every Mac user of Google products.

Tip: Use contexts in Toodledo, Appigo Todo.app and others to identify quick tasks

For historic reasons, and because between the two of them I have a unique set of features I can't find elsewhere, I use Toodledo online and Appigo's Todo.app on my iPhone (they sync). The combination is a bit of a kludge, but by now they are burned into my brain. It would take real improvements for me to switch, and I haven't seen them [1]. For example, Appigo's online task app is missing Toodledo's essential mailto feature.

Despite years of use however, I've not done much with Contexts in either Toodledo or Todo [2]. I understand the theory, but in practice it's been too much bother -- until today.

Today I realized what I needed was an easy way to tag small tasks that I can easily knock off when I have a free minute or less -- and I realized a single letter context was a convenient way to do that. So I deleted all the unused Contexts I'd created, and added a new one called 'q' (for quick!). Now, when I'm bored, I filter on q and knock off a bunch of little tasks.

[1] I could build such a great task app if I had another life or two ...
[2] Todo syncs contexts with toodledo. 

Aperture slideshow smaller than expected? You may need to create previews.

When I began using Mountain Lion the slideshow screen saver couldn't handle nested photo folder. That regression from the best version of OS X was fixed in 10.8.3, but by then I'd switched to using Aperture for my slide show of child-programming images [1].

Aperture came with its own problems. For example, the same image would seem to get stuck on both monitors. That improved with 10.8.3 and/or running 'repair database'. Some time later though, it occurred to me I was seeing the same images more often than expected for a library of 6,000 "best of" images.

The catch, of course, was previews. The slide show feature needs JPEG images to work, and, as it turned out, only a small number of my 6000 slideshow "RAW" images had previews. 

The Aperture 3 User Manual explains how to update or create previews: "Choose Maintain Previews For Project from the Library Action pop-up menu". I created a smart folder to hold all my slideshow images, chose select all then "maintain preview" on one context menu. I came back a few hours later and now I have a much larger set to view. 

[1] My theory is that by exposing the children to thousands of images of happy family moments, they will be programmed into thinking they had a happy childhood and thus will ensure I do not suffer long in my dotage.