Multiple accounts with the same hosting provider over SSH

It is fairly common to want to set up two SSH connections from a given machine to the same remote host. For example you might have both a work and a personal account on Github and you might want to pull and push changes from work and personal repositories on your machine.

If you just follow the instructions for setting up a single SSH connection and repeat it twice, your scond connection will probably not work because SSH on your machine will use the first account name to try to connect to the second account.

The way to fix this is to set up a .ssh/config file in which you define aliases for your two accounts, for example: github.com_work and github.com_personal. Then you use these aliases in the git config remote.origin.url in the respective repositories, for example:

git config remote.origin.url git@github.com_work:account_name_work/project_name

git config remote.origin.url git@github.com_personal:account_name_personal/project_name

For an example of the .ssh/config file see the Git FAQ.

Simula 67 as a successor to Algol 60

From "Software Engineering: As it was in 1968" by Brian Randall:

In the more rarified ALGOL world in which I moved, 1968 was something of a watershed. The IFIP ALGOL Committee (Working Group 2.1) which had been set up following the publication of the original ALGOL 60 Report, met in Munich not long after the Software Engineering Conference. A week-long debate of remarkable intensity culminated in a majority decision to approve the ALGOL 68 report that had been prepared by van Wijngaarden, Mailloux, Peck and Koster (41). In response, a renegade group (of which I was a member) produced a brief Minority Report (18), in which we stated that "it will be required from an adequate programming tool that it assists, by structure, the programmer in the most difficult aspects of his job, viz. in the reliable creation of sophisticated programs. In this respect we fail to see how the language proposed here is a significant step forward". My own recollection in fact is that several of us felt even then that the recently proposed SIMULA 67 Common Base Language (12) came closer to our ideal, and would perhaps turn out to have at least as much impact as an officially promulgated ALGOL 68.

I have known a long time (mainly from the writings of Edsger Dijkstra) about how in 1968 the Algol Committee split over the decision to approve van Wijngaarden's Algol 68 language, with a significant minority issuing their own report saying it was too complex to create reliable software with. I had always assumed that the minority would have preferred something like Wirth's Algol W, but from what Randall says above they might have actually gone for Simula 67. Now that would have been something. It would probably have brought forward the flourishing of object-oriented programming by over a decade

Haskell "done right"

There is a small portion of Java developers who dream about programming in Scala.
There is a sizable portion of Scala developers who dream about programming in Haskell.
All Haskell programmers want to program in Idris.

From IdrisTddNotes by Rober Peszek.

This tallies with my impression that Idris is Haskell "done right".

Category Theory without Objects

Our notion of category is that of [Eilenberg & Mac Lane, 1945]. We identify objects with their identity maps and we regard a diagram A--f-->B as a formula which asserts that A is the (identity map of the) domain of f and that B is the (identity map of the) codomain of f.

From the start of the PhD thesis of F. William Lawvere ( Functorial Semantics of Algebraic Theories and Some Algebraic Problems in the context of Functorial Semantics of Algebraic Theories, 1963).

Alloy Talk

Here is a good recent talk by Jay Parlar on using the Alloy Analyzer to find bugs in software designs.