Comments for So Jake Says: http://www.jakevoytko.com/blog Ye Olde Computer Science Blogge Sat, 17 May 2008 10:10:09 +0000 http://wordpress.org/?v=2.5.1 Comment on Somewhere Between Vim and Emacs There is a Great Text Editor by Naan Yaar http://www.jakevoytko.com/blog/2008/04/13/somewhere-between-vim-and-emacs-there-is-a-great-text-editor/#comment-6193 Naan Yaar Sat, 17 May 2008 04:46:01 +0000 http://www.jakevoytko.com/blog/?p=70#comment-6193 Hi, I have been using Viper in XEmacs with the "expert" level set to 5. It has worked wonderfully, allowing me to use the best of both editors. I completely agree regarding the need to have "kitchen sink" capability and modal editing. In Viper, I tend to use XEmacs keystrokes when I am in insert mode and use vi keystrokes for mostly everything else. The save and undo key combinations in XEmacs come in handy and I tend to use them a lot. Also, dabbrev-complete is unbelievably useful for any editing; I cannot live without this. Why do you believe Viper is not the way to go? Hi,

I have been using Viper in XEmacs with the “expert” level set to 5. It has worked wonderfully, allowing me to use the best of both editors. I completely agree regarding the need to have “kitchen sink” capability and modal editing.

In Viper, I tend to use XEmacs keystrokes when I am in insert mode and use vi keystrokes for mostly everything else. The save and undo key combinations in XEmacs come in handy and I tend to use them a lot. Also, dabbrev-complete is unbelievably useful for any editing; I cannot live without this.

Why do you believe Viper is not the way to go?

]]> Comment on How Reddit Will (Maybe) Save Software Development by Bryce http://www.jakevoytko.com/blog/2008/05/12/how-reddit-will-maybe-save-software-development/#comment-5975 Bryce Tue, 13 May 2008 01:16:51 +0000 http://www.jakevoytko.com/blog/?p=78#comment-5975 Ooooh there are flames in these here comments!!! Lol that's sadly one of the best compliments the internet can give! Seriously thought provoking article though. Sadly I'm too fried to pop off a quality comment but an enjoyable read none the less. Ooooh there are flames in these here comments!!! Lol that’s sadly one of the best compliments the internet can give!

Seriously thought provoking article though. Sadly I’m too fried to pop off a quality comment but an enjoyable read none the less.

]]> Comment on How Reddit Will (Maybe) Save Software Development by William Bland http://www.jakevoytko.com/blog/2008/05/12/how-reddit-will-maybe-save-software-development/#comment-5955 William Bland Mon, 12 May 2008 15:45:33 +0000 http://www.jakevoytko.com/blog/?p=78#comment-5955 "Theorems that have withstood the test of time are falling, one by one; just ask Fermat and Poincaré." You mean "conjectures", or "open problems" rather than "theorems". Theorems are the end-product of mathematics. Fermat's Last Theorem wasn't actually *known* to be a theorem until a proof was found - until then it was a conjecture. “Theorems that have withstood the test of time are falling, one by one; just ask Fermat and Poincaré.”

You mean “conjectures”, or “open problems” rather than “theorems”. Theorems are the end-product of mathematics. Fermat’s Last Theorem wasn’t actually *known* to be a theorem until a proof was found - until then it was a conjecture.

]]> Comment on How Reddit Will (Maybe) Save Software Development by Daniel Schumacher http://www.jakevoytko.com/blog/2008/05/12/how-reddit-will-maybe-save-software-development/#comment-5952 Daniel Schumacher Mon, 12 May 2008 15:00:31 +0000 http://www.jakevoytko.com/blog/?p=78#comment-5952 I agree with much of the setup, not the conclusion. I agree that a social methodology of *design* is needed. The software industry is immature... similar to when the auto industry made the model-T Ford (it could be argued we are earlier than that... after all the Model-T was reliable, cheap, and produced efficiently.) My solution is to create a social requirements engineering system (this would also supersede bug dbs... since a bug is merely an anti-requirement.) I would integrate this requirement system directly with the testing system (could be unit tests or other types.) I'd call this "requirement verification". This would make the "big triangle" of design-implement-test an actual triangle! Today, at software companies I've worked at the triangle's sides are only vaguely related/connected. I bought the domain "OpenRequirements.org" hoping to advance this idea... if anybody wants to work on it with me please get in touch. I do a lot of CAD design (I create CAD software... IntelliCAD.) Note that blueprint-style design is actually a thousands of years old practice. The software industry has no analogous design methodology. We are at the phase where we construct a wooden Indian from toothpicks, glue and chisels (and we stop when we have a wooden Indian or we run out of time.) It should not be surprising that our end-products are of such low-quality, given this methodology. The keys to improving our design capacity are: - Design captures requirements adequately. Social "marketplace of ideas" methods ensure this. - Design is verifiable, and verified (integrated testing.) This is like having "machine readable specifications" I agree with much of the setup, not the conclusion. I agree that a social methodology of *design* is needed. The software industry is immature… similar to when the auto industry made the model-T Ford (it could be argued we are earlier than that… after all the Model-T was reliable, cheap, and produced efficiently.) My solution is to create a social requirements engineering system (this would also supersede bug dbs… since a bug is merely an anti-requirement.) I would integrate this requirement system directly with the testing system (could be unit tests or other types.) I’d call this “requirement verification”. This would make the “big triangle” of design-implement-test an actual triangle! Today, at software companies I’ve worked at the triangle’s sides are only vaguely related/connected. I bought the domain “OpenRequirements.org” hoping to advance this idea… if anybody wants to work on it with me please get in touch. I do a lot of CAD design (I create CAD software… IntelliCAD.) Note that blueprint-style design is actually a thousands of years old practice. The software industry has no analogous design methodology. We are at the phase where we construct a wooden Indian from toothpicks, glue and chisels (and we stop when we have a wooden Indian or we run out of time.) It should not be surprising that our end-products are of such low-quality, given this methodology. The keys to improving our design capacity are:

- Design captures requirements adequately. Social “marketplace of ideas” methods ensure this.

- Design is verifiable, and verified (integrated testing.) This is like having “machine readable specifications”

]]> Comment on How Reddit Will (Maybe) Save Software Development by Greg Graham http://www.jakevoytko.com/blog/2008/05/12/how-reddit-will-maybe-save-software-development/#comment-5949 Greg Graham Mon, 12 May 2008 14:25:10 +0000 http://www.jakevoytko.com/blog/?p=78#comment-5949 When I saw you talking about Reddit and peer-review, I thought you were going to propose that programmers post their code in a Reddit-like forum, subject to votes and comments. The problem would be getting people to actually rigorously review the code. A lot of people comment and vote on things on Reddit without carefully reading the article or thinking about the subject. Reading and understanding code is more difficult than reading and understanding most articles. When I saw you talking about Reddit and peer-review, I thought you were going to propose that programmers post their code in a Reddit-like forum, subject to votes and comments. The problem would be getting people to actually rigorously review the code. A lot of people comment and vote on things on Reddit without carefully reading the article or thinking about the subject. Reading and understanding code is more difficult than reading and understanding most articles.

]]> Comment on How Reddit Will (Maybe) Save Software Development by Jake http://www.jakevoytko.com/blog/2008/05/12/how-reddit-will-maybe-save-software-development/#comment-5947 Jake Mon, 12 May 2008 13:43:35 +0000 http://www.jakevoytko.com/blog/?p=78#comment-5947 @Mitch: I tried to hint at this (and apparently didn't state the idea clear enough), but design would be another thing that this system could handle.. though design tends to be more of an abstract concept. I don't immediately see how to peer-review sections of design in the general case, but I'll certainly think about it. @Mitch:

I tried to hint at this (and apparently didn’t state the idea clear enough), but design would be another thing that this system could handle.. though design tends to be more of an abstract concept. I don’t immediately see how to peer-review sections of design in the general case, but I’ll certainly think about it.

]]> Comment on How Reddit Will (Maybe) Save Software Development by Mitch http://www.jakevoytko.com/blog/2008/05/12/how-reddit-will-maybe-save-software-development/#comment-5944 Mitch Mon, 12 May 2008 13:39:50 +0000 http://www.jakevoytko.com/blog/?p=78#comment-5944 Hey Jake, great article, I agree with you on the Truth about Software Engineering: http://softwareindustrialization.com/TheTruthAboutSoftwareEngineering.aspx However, I disagree with your approach. Write more unit tests? This is after the fact that the software has already been written – it’s too late and a) it has already cost way too much to get there, as you have already pointed out and b) the unit tests may prove (likely) that the software was not designed correctly in the first place and has to be rewritten, again at an exorbitant amount (see a). The key word I believe is design. And if you know about Jack Reeves and the source code is the design belief, then you know in reality, our world of software engineering suffers from no way to specify a software design and prove the software design works, before a single line of code has been written. I believe that is the right problem to be solving as has been done in other “engineering disciplines. Cheers! Hey Jake, great article, I agree with you on the Truth about Software Engineering:

http://softwareindustrialization.com/TheTruthAboutSoftwareEngineering.aspx

However, I disagree with your approach. Write more unit tests? This is after the fact that the software has already been written – it’s too late and a) it has already cost way too much to get there, as you have already pointed out and b) the unit tests may prove (likely) that the software was not designed correctly in the first place and has to be rewritten, again at an exorbitant amount (see a).

The key word I believe is design. And if you know about Jack Reeves and the source code is the design belief, then you know in reality, our world of software engineering suffers from no way to specify a software design and prove the software design works, before a single line of code has been written. I believe that is the right problem to be solving as has been done in other “engineering disciplines.
Cheers!

]]> Comment on How Reddit Will (Maybe) Save Software Development by Jake http://www.jakevoytko.com/blog/2008/05/12/how-reddit-will-maybe-save-software-development/#comment-5943 Jake Mon, 12 May 2008 13:06:50 +0000 http://www.jakevoytko.com/blog/?p=78#comment-5943 @Bob: I wondered if someone was going to catch that! @Bob:

I wondered if someone was going to catch that!

]]> Comment on How Reddit Will (Maybe) Save Software Development by Bob Holness http://www.jakevoytko.com/blog/2008/05/12/how-reddit-will-maybe-save-software-development/#comment-5941 Bob Holness Mon, 12 May 2008 12:45:39 +0000 http://www.jakevoytko.com/blog/?p=78#comment-5941 "...projects are still failing at an alarming rate." "Programming failures are due to a lack of rigor at all levels" "but if someone else wants to hack up their own version, awesome!" Ha ha. “…projects are still failing at an alarming rate.”
“Programming failures are due to a lack of rigor at all levels”
“but if someone else wants to hack up their own version, awesome!”

Ha ha.

]]> Comment on How Reddit Will (Maybe) Save Software Development by Dan http://www.jakevoytko.com/blog/2008/05/12/how-reddit-will-maybe-save-software-development/#comment-5939 Dan Mon, 12 May 2008 12:21:17 +0000 http://www.jakevoytko.com/blog/?p=78#comment-5939 There is a plug-in for Hudson that does something similar (although without the modding up and down): http://hudson.gotdns.com/wiki/display/HUDSON/The+Continuous+Integration+Game+plugin There is a plug-in for Hudson that does something similar (although without the modding up and down):

http://hudson.gotdns.com/wiki/display/HUDSON/The+Continuous+Integration+Game+plugin

]]>