alierak

kareila retweeted a link to this article which offers some hints of the (female) history of programming. Interesting, but unsatisfying in lack of detail. A commenter pointed out the sentence "For example, Betty Holbertson convinced skeptical engineers to include a “stop instruction” in order to guard against human error." as particularly in need of exposition. I agree. That kind of unexplained assertion is like velcro to the curious mind.

This stuff is not entirely beyond the reach of Google, I hope, but it's just not taught alongside computer science as far as I know. Here's what I found after a few minutes of poking around. I don't see how to comment on the article above.

• It's Holberton, not Holbertson (and, I think, Snyder at the time of the ENIAC work).


• Computer Oral History Collection, 1969-1973, 1977 Jean J. Bartik and Frances E. (Betty) Snyder Holberton Interview, April 27, 1973, Archives Center, National Museum of American History, page 78:
  
  HOLBERTON:
  
  It's the only time I ever talked with Von Neumann and I felt rather bold when I got up and said that, too.
  
  Well, they had the instruction code and it didn't have a stop instruction and he said, "Well, you don't need it, there are all these extra leads and you've got 99 possible combinations and it will go to any one of these grounds." And I was bold and I remember that I felt bold at the time and I said, "But we are not all Von Neumann’s, we will make mistakes and it will be unintentional." And he just nodded his head and said, "All right."
  


• That's a long transcript, but well worth reading more of. See page 56 for an etymology of "breakpoint", for example.


• There are many more records like it, easily obtained through a network of computing systems developed not entirely by men.

I'm impressed; DW actually made an appearance (however small) in xkcd.

Testing crossposter with wrong password stored in DW prefs.

This post contains a literal right single quotation mark: ’
Unicode: 2019
UTF8: E2 80 99

The sudafed woke me up at 3am today so I started looking at the directory search bug. Read some code, poked the directory search and results pages a bit, but no definite progress. I probably need to look under the hood while a search is running to get anywhere. Debug logging, figuring out whether it's using Gearman or not, etc. So I doubt I'll be of much use without a working dreamwidth installation to mess with. Hrm.

Edit: Oh, the directory geographic stats do work, so you can find out how many accounts are where. But of course that uses a different set of queries.

Edit 2: I have a theory now, at least. From UserSearch.xs comes this gem:


// FIXME: this will kinda blow w/ 64-bit pointers... we should/could just store 32-bit id instead.
resultset[resultset_size++] = &usermeta[uid];


Gee, I wonder what OS dreamwidth is running on...

Edit 3: That comment was a total red herring. Left a real diagnosis in the comments of the bug linked above. I really would have gotten nowhere fast without the hack space Janine mentioned and Sophie provided, plus this comment over on lj_dev.