I Still Know What You Learned Last Summer

Reducing merge headaches: git meets diff3

git has an option to display merge conflicts in diff3 format (by default it only displays the two files to be merged). You can enable it like so:

$ git config --global merge.conflictstyle diff3

Now, when you have to resolve merge conflicts, git shows your side, the side being merged, and (here's what's new) the common ancestor in between them. Here's an example of the diff3-formatted output:

cauliflower <<<<<<< HEAD peas potatoes ||||||| merged common ancestors peas ======= >>>>>>> topic tomatoes

Having the merge ancestor readily available helps you to quickly determine what the correct merge is, since you can infer from it the changes that were made on both sides. Here you can see that the original state was peas. On your branch potatoes was added (compare the middle section to the top) and on the other branch peas was removed (compare the middle section to the bottom). Therefore the correct change is to both add potatoes and remove peas, leaving you with just potatoes in the conflicted section.

There's really no reason you shouldn't enable the diff3 style, because you frequently need the ancestor to determine what the correct merge is.

To see that this is true, even in the simple example above, look at what the conflict looks like under the standard style:

cauliflower <<<<<<< HEAD peas potatoes ======= >>>>>>> topic tomatoes

There's an asymmetry between peas and potatoes: one was added and one was deleted, but this merge conflict doesn't tell you anything at all about which was which! You can't determine the correct merge unless you remember the sequence of changes that led up to this point. And why should you have to rack your brain to do that? That's exactly the sort of thing that your computer can, and should, help you with.

Bonus tip: rerere (reuse recorded resolution)

If your workflow finds you redoing the same merges over and over again you might also find git's rerere (reuse recorded resolution) feature to be useful.

One of the things that is wonderful about rerere is that it provides hardly any UI surface at all. Just set it...

$ git config --global rerere.enabled 1

...and forget it. Although there is a git rerere command, you can get a lot done without using it at all.

After enabling rerere, whenever you resolve a merge conflict, git automatically squirrels away the resolution in its database. You'll see a message like this one:

$ git commit Recorded resolution for 'soup' [...]

And the next time you encounter the same conflict, where you would have expected git to spit out a file with conflict markers, you will instead find that it has automatically resolved the merge for you, and printed the following message:

$ git merge topic Auto-merging soup CONFLICT (content): Merge conflict in soup Resolved 'soup' using previous resolution.

Just double-check to make sure nothing has gone awry, add, and commit. Save your blood, sweat, and tears for other, more interesting problems than redoing merges.

Making your own page-a-day calendar, revisited

Some time ago I posted instructions for designing and producing a page-a-day calendar, which is a moderately neat project at the intersection of metaprogramming and handicraft.

Matt Johnson made one for the year 2011; his write-up fills a lot of the gaps in my rather skeletal instructions and provides a number of suggestions for extending the original design in very good ways (including adding page content not based on photographs, and making a wooden stand for the calendar). Read it if you are thinking of making one of these things.

Thanks, Matt!

CC-BY Matt Johnson

2010 in Review

These were the most linked-to posts on my blogs in the last year:

Happy new year, peace on earth, and may all your builds be green!

Making the leap to SSD...

The Intel SSDs were on sale around Thanksgiving, so I picked up one for my Thinkpad.

There is little that was remarkable in getting it set up. I swapped out the old spinning rust device. (The X201, unlike any other laptop I've opened up, has rubber shock absorbers that fit between the drive and the chassis. Not that the thing they're guarding is particularly susceptible to mechanical damage any longer.) Everything just worked.

Boot times and application load times are dramatically improved. Though, the way I work, my laptop is essentially just a device for running Emacs, Chrome, and gnome-terminal, all of which I just open and keep open. All of those apps load fairly quickly, so the speed improvement from the SSD, while nice, is not by any means life-changing. (And, how often do you reboot a computer, anyway? But, for the record, my Thinkpad goes from LUKS password prompt to login screen in 10 seconds, and to a ready-to-use desktop in about 3 more.)

What's really nice about the SSD is that my laptop is now nearly silent, or at least quieter than ambient noise. Perhaps I just have unusually low noise tolerance, but I find the steady-noise of a HDD to be somewhat annoying, and the sound of a HDD spinning up to be even more annoying. Unfortunately, I actually didn't correctly attribute those noises to the HDD until recently. If this sounds like you, it's time to pay a visit to the computer store.

One caveat is that if you use LUKS full disk encryption (and if you're using a laptop, you really ought to; no, I mean, you really ought to), then Ubuntu (10.10) doesn't issue TRIM commands to the device. Which will eventually lead to decreased throughput, but I believe that may just be the price of security. (IANA security expert...) LUKS goes to great lengths by default to foil cryptanalysis, including initializing encrypted partitions with random data to obscure which sectors of the device are actually being filled with interesting data. This would all be for naught if the OS was periodically telling the device exactly which sectors were no longer being used.

Content-aware image resizing by seam carving

This is a classic, or, as much of a classic as a 3-year-old graphics paper can be.

Seam Carving for Content-Aware Image Resizing (Avidan and Shamir, 2007) describes a technique for resizing photographs along one dimension in a way that allows the aspect ratio to be changed while reducing the amount of distortion in perceptually important parts of the image. Watch the video, if you haven't seen this before:

Youtube link

The algorithm translates into identifying the shortest path in a DAG and it can be easily implemented by a dynamic programming algorithm. As Dmitry pointed out to me some years ago, this makes an ideal teaching example for dynamic programming, for a number of reasons. The algorithm is easily visualized because the space of values to be computed maps straightforwardly onto the pixels of the image. Both the naive exponential-time algorithm (testing all 3^#rows paths) and the dynamic programming algorithm are fairly easily expressed without too much additional machinery. And, unlike most classroom dynamic programming questions I have encountered, which seemed either banal or contrived, the applications of this one are visually appealing and fairly interesting. For example, as someone pointed out to me, it makes the fallout of breaking up much easier to deal with.

Before... and after (Source: video, above)

Some Android tips

I've been using a Nexus One as my primary phone for about a year now. Here are some tips and suggestions I've accumulated. Most of these tips should also apply to the Nexus S and the G2 (which have stock Android builds), as well as to other Android phones (Froyo+) modulo any vendor customizations.

Press and hold the Search button to activate voice search / voice actions. This is indispensable! I use it a least a couple of times every day. It makes the phone feel like future tech, not least because it's dramatically better than anything I regularly deal with in phone systems or on PCs. The phone recognizes special instructions, including things like the following (for anything it doesn't recognize as an action, it asks you to disambiguate or it falls back to a Google search):

"Call Phil Sung, mobile"
"Navigate to Fry's Electronics"
"Map of gas stations"
"Note to self, buy more envelopes" (which sends you an email from yourself)

Press and hold the Home button to easily get a list of recently used apps.
Press and hold the Menu button to toggle the display of the soft keyboard. Not usually needed, but occasionally useful.
When using the soft keyboard, you can drag your finger past the top edge of the keyboard for quick access to digits and punctuation.
If you add your email address to the user dictionary, you can get it in the list of autocompletions when filling out forms in the web browser.
It's important to control the quantity of notifications so that they're at a level that is actually useful. Depending on how important some event is, you can configure the phone to make noise or merely to show you a notification the next time you turn on the phone. As an example, here's how I'm set up:

My phone only rings for phone calls, SMS, and IMs.
GMail generates a notification in the notification area but no noise (configurable in GMail settings). Even so, I implemented an elaborate system of GMail filters to keep all but the most important emails from getting to my inbox and thus generating notifications.
Calendar events generate neither noise nor notifications (configurable in Calendar settings). Instead, I put the Calendar widget on my home screen, and I just look at it whenever I need to. My calendar time is pretty sparsely scheduled, though. YMMV.

Easy wifi autoconfiguration with barcodes

ZXing's Barcode Scanner Android app (the most popular barcode app for Android; it's available on the Market) can now configure a wifi network based on information encoded in a QR code.

So you can say goodbye to having guests/patrons/visitors ask you how to log in to your wifi network every time someone new visits. Instead of having to select the correct network and type in the password (I thought computers were supposed to relieve us of this kind of drudgery), all they have to do is scan a barcode, and boom, they're online.

ZXing's QR Code Generator will help you make such a barcode (select "Wifi network" in the dropdown), after which you can just print it out and leave it in the living room.

If you wish, you can also create such a barcode yourself just by encoding the SSID and password in the barcode payload. Here are a couple of examples that illustrate the encoding method:

WEP: WIFI:S:mynetwork;T:WEP;P:00deadbeef;;

(Example using Google Chart API)

WPA: WIFI:S:mynetwork;T:WPA;P:mypassword;;

(Example using Google Chart API)

Hat tip to Vikram Aggarwal who implemented this concept, originally as a standalone app, WyScan, and then later implemented it in the Barcode Scanner app.