| |||||||||||||
An LCA 2012 summaryAn LCA 2012 summaryPosted Jan 26, 2012 13:26 UTC (Thu) by etienne (subscriber, #25256)Parent article: An LCA 2012 summary
In my experience strings (error messages, menu, ...) is a big part of a well written application, and if you try to support 10 to 50 languages the software size explodes.
(Log in to post comments)
An LCA 2012 summary Posted Jan 26, 2012 15:59 UTC (Thu) by cesarb (subscriber, #6266) [Link]
Strings for other languages do not count, since they are stored in separate files.
Of course, the infrastructure responsible for loading these separate files still counts.
An LCA 2012 summary Posted Jan 27, 2012 15:50 UTC (Fri) by jeremiah (subscriber, #1221) [Link]
I have some of these issues with postgres. Because I can get away with it, I always run my servers with LANG=C. The minute you use a different LANG postgres's index speeds go to crap. It's just not worth the cost when all of your data is in US-en to begin with. ASCII was good enough for grandpa, it should be good enough for me!
programs growing in size over the years Posted Jan 27, 2012 19:58 UTC (Fri) by giraffedata (subscriber, #1954) [Link] Considering just the base language, I think it's true that in Unix 6 programmers consciously avoided detailed messages in order to save the space to store the text. To the extent that the new larger programs are that way because the error messages are more informative, we definitely got our money's worth. But it looks to me like the three word error message is still king in the Unix world and messages longer than one line are rare. So is having separate messages for separate conditions. So I don't think better messages accounts for much of the bloat.
|
|||||||||||||