Date-Preserving Patch Folding with Mercurial Queues

Mercurial Queues are a powerful tool for organizing changes into logically coherent commits. My current workflow often goes something like this:

1. Make some changes implementing feature A (as a patch).
2. Generate a series of patches moving on to feature B.
3. Realize, after working on feature B, that feature A needs a small tweak, or a better commit message.

The relevant mq commands will get the job done in a jiffy. However, there’s a slight downside: after all that qpushing and qpopping, the patches in the series will lose their original dates, which means that qfinish will produce a tightly clustered series of patches that don’t actually reflect the true chronology of their development.

Fortunately, Mercurial provides the tools needed to preserve a meaningful chronological order. We’ll exploit hg log’s configurable formatting to make it generate a shell script that’ll set the dates for our patch series automatically. A pretty little hack, no? Here are the relevant incantations, assuming you want to modify a patch named P with revision number N:

1. hg log -r[[N]]:tip --template "hg qref -d '{date|date}'; hg qpush;\n" > d.sh
2. Run hg qgoto [[P]], then make whatever changes you need.
3. Run bash d.sh, which will restore the original dates for all patches in the queue following the modified one.

An Apple Easter Egg, Of Sorts

Looks like someone at Apple has a bit of a sense of humor:

 

 sosumi for the legal section, eh?

Apparently they did the same thing for the 10 billion download promotion, too.

Agrajag Rediscovered

Going through the files on an old family computer, I discovered a long-lost email exchange with Douglas Adams:

Subject: Re: Agrajag
Date: Fri, 11 Jun 1999 17:52:48 -0700
From: "Douglas Adams" <askdna@TDV.COM>
To: Martin Karel <karel@BELLATLANTIC.NET>

Irene –
I’m pretty bad at planning what’s going to happen next week. Alex is exactly right.

Best,
———————————————————–
Douglas Adams | <http: http://www.douglasadams.com>;
———————————————————–
The Digital Village | <http: http://www.h2g2.com>;
———————————————————–

Please – help save the last 650 mountain gorillas
http://www.gorillas.org/index.html

> Hi,
>
> Please settle a debate between me and my brother. I held forth that
> you must have already had Agrajag in mind when you wrote the bit about
> the pot of flowers that, when falling, said "Oh no, not again!" But
> Alex thinks that you probably just randomly stuck that line in there,
> and came up with Agrajag later and it just happened to fit. If I’m
> right, then we are very impressed with your degree of planning ahead,
> seeing as how it was four years between the publishing dates of
> "Hitchhiker’s Guide" and "Life, the Universe, and Everything."
>
>            - Irene Karel,
>               an adoring fan

Perhaps worth noting: the saved text file was 1337 bytes.

Firefox Add-On Compatibility

It looks like Firefox is moving to a compatible-by-default model for extensions. I find this interesting, since I proposed this back in 2004 to no avail.

To be clear, I don’t mean to say that compatible-by-default would have been the right choice in 2004… Predictions about the past are almost hard as predictions about the future, and at least two major factors have changed since then. First, the state of the world for extensions now is more stable than it was in the Firefox 1.0 timeframe. And second, Firefox’s rapid release cycle exacerbates spurious compatibility problems, making this proposal significantly more attractive.

The Lessons of Lucasfilm’s Habitat

We were initially our own worst enemies in this undertaking, victims of a way of thinking to which we engineers are dangerously susceptible. This way of thinking is characterized by the conceit that all things may be planned in advance and then directly implemented according to the plan’s detailed specification. For persons schooled in the design and construction of systems based on simple, well-defined and well-understood foundation principles, this is a natural attitude to have.

http://www.fudco.com/chip/lessons.html

I wonder what Intel is up to?

 

http://cufp.galois.com/2007/slides/AnwarGhuloum.pdf

“A proposed long term solution: a new functional language that features implicit parallelism, dependent typing, and an effects type system”

“We are also working with an ISV on a compiler for a parallel functional language
- Strict, but not call by value for implicit parallelism (maybe
with lightweight annotations)
- Arrays and array comprehensions for data parallelism
- Effects system to contain impure features
- Atomic for state updates”

 

http://cufp.org/conference/sessions/2010/functional-language-compiler-experiences-intel

“For five years Intel’s Programming Systems Lab (PSL) has been collaborating with an external partner on a new functional programming language designed for productivity on many-core processors. While the language is not yet public, this talk outlines motivations behind the language and describes our experiences in implementing it using a variety of functional languages. The reference interpreter is written in Haskell and compiled with GHC while PSL’s performance implementation is written in SML and compiled with Mlton. We have also generated Scheme code compiled with PLT Scheme as part of a prototyping effort.”

A Small Example Illustrating Why Unification-Based Type Inference Is Not Always The Most User-Friendly Choice

let base = 10
let bitsPerDigit = logBase 2.0 (fromIntegral base)
let activeBits = fromIntegral $
  ceiling (bitsPerDigit * (Prelude.length "123"))

There's a missing fromIntegral above, which causes GHC to underline logBase and complain that it has no instance for (Floating Int). This is, on the face of things, completely mystifying. Both parameters are “obviously” floating! A second error points to ceiling, saying no instance for (RealFrac Int).

The second error is the more suggestive one, since the problem is that the type of * causes GHC to equate the type of bitsPerDigit with that of (Prelude.length clean), i.e. Int.  But it’s more than a little strange to force the programmer to reason about their code in the same order that the type checker does!