Bug #2736
Mis à jour par Alexey Lavrentev il y a plus de 5 ans
TXM version: 0.8.0.202001161703 org.txm.rcp.app null
OS: Ubuntu 16.04
h1. To reproduce the bug:
Load ORTOFON2ANNOT3 binary corpus from https://sharedocs.huma-num.fr/wl/?id=vjA7XIdgjw5YCpVVJ1QJIAxQNYmDpTld https://sharedocs.huma-num.fr/wl/?id=NFHkyBWoAFkiEZnlNxbEU6Ai9FOZgOCN (created with TXM 0.7.9 XTZ module)!
Make a concordance of any word (e.g. "voda")
Double-click on a concordance line.
h1. Diagnostics
The problem is due to the editions generated with XSLT. The txm:first-word-id meta is set to "w_0" for the first (title) page of every text. TXM 0.7.X first retrieved the text, then the page, so for 1-page editions the right page was displayed. TXM 0.8.0 looks directly for word identifiers.
h1. Solution
h2. Step 1
Change the XSL producing the editions to set the first-word-id to w_textId_0 for the first page.
h2. Step 2
When loading a 0.7.9 corpus check the first-word-id of the first page and insert text identifier if necessary
h2. Step 3
Work out a more robust back-to-text mechanism.
OS: Ubuntu 16.04
h1. To reproduce the bug:
Load ORTOFON2ANNOT3 binary corpus from https://sharedocs.huma-num.fr/wl/?id=vjA7XIdgjw5YCpVVJ1QJIAxQNYmDpTld https://sharedocs.huma-num.fr/wl/?id=NFHkyBWoAFkiEZnlNxbEU6Ai9FOZgOCN (created with TXM 0.7.9 XTZ module)!
Make a concordance of any word (e.g. "voda")
Double-click on a concordance line.
h1. Diagnostics
The problem is due to the editions generated with XSLT. The txm:first-word-id meta is set to "w_0" for the first (title) page of every text. TXM 0.7.X first retrieved the text, then the page, so for 1-page editions the right page was displayed. TXM 0.8.0 looks directly for word identifiers.
h1. Solution
h2. Step 1
Change the XSL producing the editions to set the first-word-id to w_textId_0 for the first page.
h2. Step 2
When loading a 0.7.9 corpus check the first-word-id of the first page and insert text identifier if necessary
h2. Step 3
Work out a more robust back-to-text mechanism.