On Thu, Nov 07, 2013 at 10:28:50AM +0100, Andras Timar wrote: > Hi Khaled, > > On Wed, Nov 6, 2013 at 4:23 PM, Khaled Hosny <khaledho...@eglug.org> wrote: > >> Well, we could have avoided re-translating the 470 words in shared, > >> because the translatable strings have not changed, only identifiers. > >> That's a pity. If only Pootle were more intelligent, and distinguished > >> tags from text... > > > > Can you elaborate on this issue? > > > > For example: > > Old English string: > <ahelp hid="HID_ALIGNMENT">Sets the alignment options for the contents > of the current cell, or the selected cells.</ahelp> > > New English string: > <ahelp hid="cui/ui/cellalignment/CellsAlignPage">Sets the alignment > options for the contents of the current cell, or the selected > cells.</ahelp> > > Only the hid attribute of ahelp tag changed, which is not localizable > anyway. Yet, upon update the old translation is not even offered as > fuzzy, because fuzzy matching algorithm does not distinguish tags from > text. Tags in segments are usually invariable.
Is the <ahelp> tag almost alway a top level one? If yes, then I think the best way is for PO extraction tool to strip the tags completely and use the hid as msgctx (the PO to help converter will have to be updated, too). > It would make sense to take this into account, when you build a local > TM and/or eveluate matches from the local TM. I think the fuzzy matching in translate toolkit is pretty simplistic, it probably uses Levenshtein distance and don't do any parsing of the text. We have some ideas on how to improve this, but nothing implemented yet. You can try to experiment with pomigrate2 tool from translate toolkit with or without -p option and see if it gives you better results (with -p option it will use the same logic used by Pootle). Regards, Khaled -- To unsubscribe e-mail to: l10n+unsubscr...@global.libreoffice.org Problems? http://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/ Posting guidelines + more: http://wiki.documentfoundation.org/Netiquette List archive: http://listarchives.libreoffice.org/global/l10n/ All messages sent to this list will be publicly archived and cannot be deleted
