Cher Maca,

( je ne sais plus d'où je tiens cela dans un coin de ma tête ! peut-être même pas exact
)
Merci pour la réponse.
J'ai vite ajouté un nouveau
LCS6(20000 + 1, 20000 +1 ) as integer
et cela tient. Il me reste encore de la mémoire.
En outre, je dois garder as integer, car le tableau comprendra les occurances des enregistrements des deux fichiers chargés en mémoire.
Comme j'ai dit il s'agit de traiter une séquence LCS dont le pseudo code est
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
|
# Find the index of the first mismatch top down
for i = 1 to |F|
if F[i] = F[i] then continue
if F[i] <> F[i] then break
next
# Files are identical
if i = |F| then return TextSpan(1,0)
# Find the index of the first mismatch bottom up
# Redefine F[j] to mean the 𝑗
𝑡 character up from the bottom line
for j = |F| to i
if F[j] = F[j] then contine
if F[j] <> F[j] then break
next
if (j - i) > k then
# Compile the entire file, they are too different
return Textspan(1, |F|)
else
# The change occurred between i and j
return Textspan(i, j)
end if |
( exemple de Final Project Applied Algorithms - CSEP 521 - Chris Smith and Meher Malakapalli )
WinDiff chez moi parvient à traiter deux fichiers de 700.000 enregistrements chacun.
Je suppose qu'il fait cela en mémoire ( vu la rapidité du traitement ).
Moi je n'y arrive pas .............. encore 

Je vais donc plancher ....
Bonne soirée et encore merci
Pierre
Partager