CommonsDelinker

Revision as of 10:44, 11 March 2021

Revision as of 13:02, 11 March 2021

Line 311:

:{{ping|SlowPhoton}} It was (probably) me, but it was supposed to be fixed afterwards. In a nutshell: there is no maintainer for the bot as of now, so I am the one trying to fix the problems which caused the bot to be stuck for months. In due course I have fixed about a dozen small problems and one of those was the replacement matching. The problem you noted may be related to this, I'll look into the batch where it happened; the problem is that it was too long ago and the bot has no logic to prevent overwriting changes inbetween. I try to create some kind of fixing run if possible, and until then I manually revert it (but it's a broken image anyway). If there are more problems please ping me here, or elsewhere. (I wrote the updates [[User_talk:CommonsDelinker#Resurrecting|there]].) Thanks! -- [[user:grin|grin]] [[user talk:grin|✎]] 10:44, 11 March 2021 (UTC)

:{{ping|SlowPhoton}} Now this is something truly fascinating (analogous with the word "trouble"). I suppose this needs some extensive cultural/literary/typographical background about Assamese writing system and its relations with the Unicode standard. What you write up there are '''not letters according to php''' (or, to be honest, it's not php to blame this time but libpcre, the matching library) and thus are not replaced as such. The problem seems to be with multiple characters: চি (due to the combining mark ি ) is not a letter according to php, and that breaks replacing; choosing a random text (বাবৰী মছজিদ) shows multiple "non-letters": ৰী and জি. Obviously I cannot fix php (or libpcre), so I have to figure out a way to avoid trusting Unicode character classification; I am not sure yet how, but I'll look into it. (Strictly there is no other ''proper'' solution: this has to be fixed…) Until then I blacklist Assamese Wikipedia in the bot. --[[user:grin|grin]] [[user talk:grin|✎]] 13:02, 11 March 2021 (UTC)