Thanks for the feedback, my comments below. On Wed, 2007-02-07 at 11:47 +0100, Christian Perrier wrote: > Talking with Frans Pop in a conference (where we indeed summarize the > results of the Extremadura sessions), he mentioned me some comments he > had pending for a while about the design and lyout of Pootle. Let's share: > > ----- Forwarded message from Frans Pop <[EMAIL PROTECTED]> ----- > > From: Frans Pop <[EMAIL PROTECTED]> > To: [email protected] > Subject: Re: Example "big" Pootle project on i18n.debian.net: debconf > translations > Date: Wed, 7 Feb 2007 11:43:52 +0100 > X-Mailing-List: <[email protected]> archive/latest/8245 > X-CRM114-Status: UNSURE (7.5861) This message is 'unsure'; please train it! > > On Wednesday 27 December 2006 09:32, Christian Perrier wrote: > > With a few workarounds and glitches, I have been able to import all > > Debian "debconf" translations in Debian's Pootle server (STILL > > EXPERIMENTAL, don't use for production work). > > I've taken a quick look and must say I'm totally confused by one thing. > > Take as example: > http://i18n.debian.net:8080/nl/debconf/clamav-data/? > > This file has 2 strings of which 1 if fuzzy. However, the translation > stats show 82% translated. WTF? > It seems that the percentage is based on a count of words, but this seems > to me a totally bogus statistic because either a whole string is > translated or it is not. > > Note that the biggest problem with a lot of translations is that people > translate words instead of the meaning of sentences. To me this statistic > seems to "promote" that :-(
Yes it is doing word counts not string counts for percentages. I would argue that string counts give absolutely no indication of the workload, they are in fact a basic binary measure (done, not done). Compare a GUI to documentation. While they could have the string count you can be assured that the docs will have at least 5 times as many words and thus be in the order of 5 times more work. Strings tell you none of that. I disagree that word counts promote word thinking vs sentence thinking. That is a fault of the translator not a statistic, you can't blame a measure for that. In the professional translation world they counts words and for some languages characters. Strings is a foreign concept. > Another comment based on: > http://i18n.debian.net:8080/nl/debconf/ > > For larger files it is not all that easy to see if there's one or two > strings untranslated or fuzzy as the grey/red area will be very small. > Would it be possible for such translations to get a different colored > folder icon? Or rather: use a different icon for files that have > ((translated+fuzzy) > 1 and (fuzzy+untranslated) > 1). That's a really good suggestion, I like that one. I'd suggest that we use the colours that are used in the progress bar. Ie Red if anything is untranslated, yellow if there is a fuzzy entry, green if everything is translated. Or at least some indicator like that. -- Dwayne Bailey Translate.org.za +27-12-460-1095 (w) +27-83-443-7114 (cell) ------------------------------------------------------------------------- Using Tomcat but need to do more? Need to support web services, security? Get stuff done quickly with pre-integrated technology to make your job easier. Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 _______________________________________________ Translate-pootle mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/translate-pootle
