Difference between revisions of "Talk:Immutable Strings"
Peter gummer (Talk | contribs) |
(Added comment) |
||
| Line 4: | Line 4: | ||
--[[User:Peter gummer|Peter gummer]] 06:00, 19 April 2007 (CEST) There may be a decent migration path that wouldn't break code. Consider the fact that Eiffel is about to embark on its ''Void-safety'' adventure: at first sight, this looks like the mother of all breaking changes, but they have a reasonable-looking plan that should avoid most of the pain. The case of mutable strings seems very similar: a common cause of runtime errors that can be fixed by reversing the current behaviour. I think it's worth considering. All of the other options, with the exploding number of variations of strings (don't forget UC_STRING, etc.) and their conversion routines look much more scary to me. | --[[User:Peter gummer|Peter gummer]] 06:00, 19 April 2007 (CEST) There may be a decent migration path that wouldn't break code. Consider the fact that Eiffel is about to embark on its ''Void-safety'' adventure: at first sight, this looks like the mother of all breaking changes, but they have a reasonable-looking plan that should avoid most of the pain. The case of mutable strings seems very similar: a common cause of runtime errors that can be fixed by reversing the current behaviour. I think it's worth considering. All of the other options, with the exploding number of variations of strings (don't forget UC_STRING, etc.) and their conversion routines look much more scary to me. | ||
| + | |||
| + | --[[User:Paulb|Paulb]] 20:48, 25 April 2007 (CEST) Starting from scratch is never an option, as Peter mentions there would actually be a migration path. We do our very best to not break code. In fact we still support every construct and keyword we've introduced in the language, even though they have been deprecated for years. | ||
Latest revision as of 09:48, 25 April 2007
Starting from scratch
While it is very galling that C# and Java should be better than Eiffel in this respect, I am afraid that starting from scratch isn't really an option. Almost all programs would be broken, and the cost of fixing millions of lines of code would be far too high.
--Peter gummer 06:00, 19 April 2007 (CEST) There may be a decent migration path that wouldn't break code. Consider the fact that Eiffel is about to embark on its Void-safety adventure: at first sight, this looks like the mother of all breaking changes, but they have a reasonable-looking plan that should avoid most of the pain. The case of mutable strings seems very similar: a common cause of runtime errors that can be fixed by reversing the current behaviour. I think it's worth considering. All of the other options, with the exploding number of variations of strings (don't forget UC_STRING, etc.) and their conversion routines look much more scary to me.
--Paulb 20:48, 25 April 2007 (CEST) Starting from scratch is never an option, as Peter mentions there would actually be a migration path. We do our very best to not break code. In fact we still support every construct and keyword we've introduced in the language, even though they have been deprecated for years.
