Difference between revisions of "Flag diacritics considered harmful"

From Apertium
Jump to navigation Jump to search
(Created page with " <pre> <Flammie> convincing arguments come through empirical evidence of real world data <Flammie> almost only flag diacritics that give sizable growth in elimination are thos...")
 
 
(One intermediate revision by one other user not shown)
Line 2: Line 2:
 
<pre>
 
<pre>
 
<Flammie> convincing arguments come through empirical evidence of real world data
 
<Flammie> convincing arguments come through empirical evidence of real world data
<Flammie> almost only flag diacritics that give sizable growth in elimination are those examples in karttunen's finnish numerals
+
<Flammie> almost only flag diacritics that give sizable growth in elimination are those examples in
  +
karttunen's finnish numerals
<Flammie> besides, flags were meant to serve two totally independent uses that have no relation, optimisation and the morphotactics, they should not be interlinked
+
<Flammie> besides, flags were meant to serve two totally independent uses that have no relation,
<Flammie> iff one of the +- tags defined in spectre's scheme actually gives any measurable growth, then we can come up with hyperminimisation scheme
 
  +
optimisation and the morphotactics, they should not be interlinked
<Flammie> having linguists write flags and think that they're doing positive impact to automaton size is as much non-sense as you'd ask python programmer to optimise code by inline assembly
 
 
<Flammie> iff one of the +- tags defined in spectre's scheme actually gives any measurable growth,
  +
then we can come up with hyperminimisation scheme
 
<Flammie> having linguists write flags and think that they're doing positive impact to automaton
  +
size is as much non-sense as you'd ask python programmer to optimise code by inline assembly
 
</pre>
 
</pre>
   
 
[[Category:Development]]
 
[[Category:Development]]
  +
[[Category:Flag diacritics]]

Latest revision as of 06:39, 20 October 2014

<Flammie> convincing arguments come through empirical evidence of real world data
<Flammie> almost only flag diacritics that give sizable growth in elimination are those examples in 
 karttunen's finnish numerals
<Flammie> besides, flags were meant to serve two totally independent uses that have no relation, 
 optimisation and the morphotactics, they should not be interlinked
<Flammie> iff one of the +- tags defined in spectre's scheme actually gives any measurable growth, 
 then we can come up with hyperminimisation scheme
<Flammie> having linguists write flags and think that they're doing positive impact to automaton 
 size is as much non-sense as you'd ask python programmer to optimise code by inline assembly