Re: [PEPr] Comment on Internationalisation::Maketext

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 > But that is YOUR personal opinion! Dear Hans, I am a linguist myself and I run a small translation company. Managing translations is my daily work. Sometimes, in my spare time, I participate in Free Software projects like pgAdmin which has more than 30 languages: http://www.pgadmin.org/pgadmin3/translation.php Maketext and Plural Gettext cannot solve the questions of languages, because languages are too complex to be coded. By "coding", I mean expressing plural, accusative, absence of number, etc... I see no problem in linguists using Maketext. The problem is that average PEAR users may be lead to think that "coding" human sentences is possible (the class it "there", so it is "works"). The idea is reinforced by the fact that most PEAR users are English speakers. In my daily work, I spend an hour daily explaining to programmer that they should write simple and complete sentences, without coding them. To me, Maketext looks like an experiment, just like plural Gettext. When plural Gettext arrived, I was one of the only persons to say "NO, it does not work". And now that everyone agrees that Plural Gettext does not work, people want to go further into misunderstanding. To be clear: 1. Coding a human sentence in a computer language is impossible. 2. When it becomes possible, it will mean that computers are intelligent. 3. This day, we can stop working and ask computers to do the job. Therefore: 1. Maketext is a research project, but it does not solve problems. 2. It can only lead PEAR users to believe that human sentences can be coded. 3. People will feel very dissapointed when they understand that it is impossible. What will a person entering 10.000 stings by hand think of Maketext in case of problem? At least, I would expect a warning in the PEAR package explaining that some translations may remain at "fuzzy" stage. You don't have to be worried in such a disclaimer, you only have to accept that computer languages cannot be used to code human languages. Cheers, Jean-Michel -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.4 (GNU/Linux) iD8DBQFAc9XtextoHHj2YFMRAsSsAKDUfcNwLjNn0wZQNZXGhVb35kvTZwCeNG90 Zav28hk8Qd2racdVq7VpKXM= =FAaY -----END PGP SIGNATURE-----