GNOME Bugzilla – Bug 696869
Different format for list of letters to allow digraphs
Last modified: 2013-08-18 15:47:38 UTC
At the moment, the system pulls the list of letters for letter games from a single line in the po where all letters are added in a row (abcdefghi...). This works ok for languages which only use monographs (like a b c d e) or precomposed complex letters (like ď ť in Slovak) but it's problematic for languages which use di or trigraphs which are not precomposed (a digraph being a combination of two letters which is seen/treated as a single letter). This affects a lot of languages - even though not all potentially regard them as individual members of the alphabet (like German sch, English ch). Nonetheless there are languages where these are handled rigidly as single units, such as Welsh ll, dd, ff or in our case (Scottish Gaelic) bh, ch, dh... and where the absence of a distinction between (for example) b (phonetically close to /b/ and bh (phonetically /v/) leaves a huge gap in the system. Not being a coder I'm not sure as to the best solution but perhaps if this data was removed from the po and provided in a file for locale data? Or perhaps if it was possible to put delimiters between the letters, e.g. a|à|b|bh|c|ch and so on?
I was not aware of this issue. I like the option to put a | separator in the po file for the alphabet. This require to rework several activities to support this feature. As I don't plan to work on this, I added your request to our idea list there: https://gcompris.net/wiki/Ideas_for_activities#Language_learning If this feature is important for you, please consider providing a patch.
It's important but I'm afraid I'm mostly a translator. I will speak to the other person working on Gaelic but I think she said she's having problem compiling it so she's not confident she can do patches. How much work is involved, if you can estimate? Maybe we can find someone else just to help with this aspect.
It is about 2 days of work.
I have never coded in C or Python before, so I will need to take a look. If somebody could volunteer to help me get the dependencies compiled on Windows, that would be great. I have already identified where the strings potentially come up in the code: src\click_on_letter-activity\resources\click_on_letter\default-en.desktop src\click_on_letter-activity\click_on_letter.c src\gletters-activity\gletters.c src\hangman-activity\hangman.py src\memory-activity\memory.c config.guess configure depcomp
You just cannot do this on Windows. It is way too complex. Much easier to do this on GNU/Linux, everything is readily available to build GCompris. Check the wiki here http://gcompris.net/wiki/Developer%27s_corner
Luckily somebody generously gave me his Linux laptop, so I'm having a go with that. I do have a Linux VM installed, but it's too slow to run any sort of IDE on it. I'll let you know if I need help.
Created attachment 240629 [details] [review] Multigraph letters for the Hangman game
Here's a patch for the Hangman game to allow for digraphs. I have also allowed the option of having a separate word list, because short words are fiendishly difficult in hangman. More on how everything works below. Please test everything before you commit. I need someone to do the following: 1. Replace the strings for the vowels and consonants in the PO file with these: msgid "a/e/i/o/u/y" msgstr "" msgid "b/c/d/f/g/h/j/k/l/m/n/p/q/r/s/t/v/w/x/z" msgstr "" 2. Add a new string for keystroke equivalences to the PO file: msgid ""a=a"" msgstr "" 3. Add the new XML files to the Makefile hangman/default-$LOCALE.xml This is how the changes work: 1. Translators now can add multigraphs to the keyboard, and the words are then parsed accordingly. For example, Germans can now have "b/c/ch/ck/d/f/g/h/j/k/l/m/n/p/q/r/s/sch/t/v/w/x/z" if they want. 2. If the person playing the game tried to type "sch", of course "s", "c" and "h" will be selected differently. So, this list can be used to define keyboard shortcuts for the multigraphs. For example, I could define sch=S, then I get "sch" if I type an uppercase S. It is also possible to have multiple shortcuts for the same multigraph. I have kept and adapted the old equivalence code (msgid "NONE", format "e=éè a=àâ"), but it doesn't seem to do anything. What is this supposed to do? I might have broken it. 3. Translators can now create separate wordlists for the hangman game. The format is identical to the format for the list in wordsgame. If there is no xml in hangman, the game looks for it in wordsgame. If still nothing is found, the game reverts to English, first looking for the English file in the Hangman folder and then finally in the wordsgame folder. I think we need to document this in the Wiki for the translators. I am still stuck on traversing strings in C, so I would be grateful if somebody could take up patching the click on letter activity.
Created attachment 240669 [details] Multigraph letters for the Hangman game I had overlooked that I was still printing some test output to the console that we don't need. So, here's an updated version.
I have committed fixes for Hangman, Wordsgame and reading-h/-v to my git fork. I still have to look into gletters, click_on_letter and memory.
All done and committed to my GIT fork. The fixes above are already in the main fork. I have also updated the Wiki accordingly.
This problem has been fixed in the development version. The fix will be available in the next major software release. Thank you for your bug report.