That book also has many other interesting papers like "The UNIX Programming Environment", "EMACS: The Extensible, Customizable, Self-Documenting, Display Editor", and "The LISP Machine"
I agree. I thought of the Do What I Mean (DWIM) part of the Xerox 1108 Lisp Machine when I first started reading this article.
Unlike some of the other commentators here, I think this is a rational approach to building an assistant pair programmer. For sceptics, I would point you research for writing code from technical papers (e.g., https://www.ibm.com/blogs/research/2018/02/deep-learning-mod...)
A blast from the past. But one which I've been thinking about again lately.
VR/AR will change IDE/PA UI constraints. Alter the economics of information display. "Screen" real estate gets cheaper. There's more opportunity for "I'm just guessing, and you probably don't actually want this, but just in case, it's here off to the side, where it won't bother you, but is available at a glance".
"Oh, you just typed an import statement for an npm module on github? Well... just in case, of to the side, here are its npmtrends, alternative modules, recent news and discussion, analysis of repo, of code, of forks, of issues, of its dev community and individuals, a performance profile from its test suite, ... [90% cache hit with week-old tolerance; $0.05 spent on Lambda/GCP for update; the following items are also available...]"
No hardware: no software. Nobody wrote Google until cheap servers were available, and nobody made antibody treatments for cancer until cheap protein synthesis became available. This paper is cute and fun but ultimately worthless intellectual masturbation.
My god this is pretentious.. Most of the actual nuts and bolts stuff they are ultimately talking about is all relatively simple known stuff (as far as I can tell from skimming)
But their choice of words, use of language, and also all the new terms and acronyms they are making up..
The authors intent certainly seems to be to confuse people and come across as intelligent to those who dont actually do any probing into the meaning of text.
Before programming, I was an architect. This has all the hallmarks of the many bullshit, unnecessarily wordy, borderline meaningless, architectural tracts one comes across in that field.
One of the reasons I got out of architecture was that is was fully permeated with bullshit, and no-one questions it.
My experience with programmers in general is for them to be much more pragmatic, direct and clear in their communication, and evidence driven.
Its sad to see that this sort of intellectually dishonest, architectural journal crap also exists in the software field.
Lets please try and keep it down.
7 comments
[ 3.1 ms ] story [ 38.9 ms ] threadInitial Report on a LISP Programmer's Apprentice https://dspace.mit.edu/handle/1721.1/6920
The Programmer's Apprentice: Knowledge Based Program Editing https://www.researchgate.net/publication/220071274_The_Progr...
Both of those papers are included in the 1984 book "Interactive Programming Environments" https://www.amazon.com/Interactive-Programming-Environments-...
That book also has many other interesting papers like "The UNIX Programming Environment", "EMACS: The Extensible, Customizable, Self-Documenting, Display Editor", and "The LISP Machine"
Unlike some of the other commentators here, I think this is a rational approach to building an assistant pair programmer. For sceptics, I would point you research for writing code from technical papers (e.g., https://www.ibm.com/blogs/research/2018/02/deep-learning-mod...)
VR/AR will change IDE/PA UI constraints. Alter the economics of information display. "Screen" real estate gets cheaper. There's more opportunity for "I'm just guessing, and you probably don't actually want this, but just in case, it's here off to the side, where it won't bother you, but is available at a glance".
"Oh, you just typed an import statement for an npm module on github? Well... just in case, of to the side, here are its npmtrends, alternative modules, recent news and discussion, analysis of repo, of code, of forks, of issues, of its dev community and individuals, a performance profile from its test suite, ... [90% cache hit with week-old tolerance; $0.05 spent on Lambda/GCP for update; the following items are also available...]"
Before programming, I was an architect. This has all the hallmarks of the many bullshit, unnecessarily wordy, borderline meaningless, architectural tracts one comes across in that field.
My experience with programmers in general is for them to be much more pragmatic, direct and clear in their communication, and evidence driven.
Its sad to see that this sort of intellectually dishonest, architectural journal crap also exists in the software field. Lets please try and keep it down.