5balls / coleitra Goto Github PK

coleitra version

unreleased

Operating System

Debian 11

Git commit

8cd4f7f

Description

I need to test this form!

Steps to reproduce

1. Got to the issue page
2. Choose the right template
3. Fill everything out

Perceived severity by user

Low

Additional information

This forms are a useful feature.

Code of Conduct

• I agree to follow coleitras's Code of Conduct

Might put this in a seperate repo for license reasons but this should be supported on both ends, so it needs to be here as well.

• Implement parser for wiktionary tables (is ok license wise because we don't use the LUA code)
• Implement grammar provider based on this parser
• Implement android service which provides grammar information from wiktionary (This convoluted arrangement is necessary for using the LUA scripts from wiktionary which are under an incompatible license).
• Implement android service which provides grammar information from other code (Find this open source code again)

Android services are something for much later. I consider the base functionality of the grammar provider working good enough for version 0.1 (the whole grammarprovider was originally planned for a later release 0.2, but we needed to implement this earlier).

There should be a better vocable editor than in old coleitra to make it usable for people other than myself.

• Find more clever way to access database singleton (could be QObject parent of edit class)
• Create good edit element for entering text
• Create expert mode which allows entering everything
• Create simple mode for easy entering
• Modify expert mode for search / editing capability
• Modify simple mode for search / editing capability

On further thought I think search / editing capabilities don't belong in those modes. Database mode is better for this and fine (should be complete though)

We need an easy way to extend the grammarprovider

• Write Qt TableView (C++) for json data structure for language
• Write QML Element for showing this as editable table
• Write interfacing functions for QML to show it from context with filled out table from editing mode (to easily switch back and forth between this mode and editing mode)

coleitra version

unreleased

Operating System

Debian 11

Git commit

No response

Description

Bug does occur always when quitting program from simple edit mode.

Steps to reproduce

1. Goto Simple Edit mode
2. Quit program
3. Observe segmentation fault

Perceived severity by user

No response

Additional information

Order of destructors is not enforced right now and m_grammar_provider is not available anymore when the destructor is called. Do we even have to disconnect this signals or should we care?

Code of Conduct

• I agree to follow coleitras's Code of Conduct

Add lua again (should be fast as it works just fine on old coleitra).

But than also think about how to elegantly replace compiled C code at runtime with LUA scripts where possible and safe. This requires exposure of the necessary functions.

Basic JSON / JSON schema support is now included and we have a first schema. Now we can implement a generic approach to work with configuration files.

• Create datastructure for json data
• Check if file for language is available in getGrammarInfoForWord and load if not already loaded
• Write generic function for parsing wiktionary pages if language file is available
• Convert static functions into configuration files for languages
• Remove static remains of language specific parsing

There should be an option to delay the grammar lookup to later. This will simplify input of vocables.

• Refactor context dependent code to be context independent
• Clean up lookup code
• Implement schedule cache option

Would be nice if it works. Not strictly necessary right now I guess

• Figure out if mingw is enough (maybe clang can do it as well?)
• See, if we can use precompiled Qt lib for windows
• See, if we can use some stupid installer (Nullsoft?). Is there something nicer for windows 11?

Android compile is probably broken right now due to new libraries. Have not done any tests to compile it on my new laptop.

• Install android SDK
• Install android NDK
• Compile OpenSSL for android (try to get it right the first time)
• Compile android variant of Qt libraries (try to get it right the first time)
• Compile nlohmann json library for android
• Compile json schema parser library for android
• Try to compile LAPACK for android
• Compile levmar for android
• Build APK package
• Document android compile in new Install chapter in coleitra.w

There are some weaknesses in the old coleitra of the database schema / the implementation of the database handling:

• Foreign keys are probably not used throughout
• Languages are not stored as a separate table but strings are used
• Only the time when a vocable leaves the buffer is stored (i.e. when it is known), but it should be stored, when a vocable is displayed and when it is not known as well
• There should be an easy way to delete all rows depending on a foreign key postponed for later
• There is still some bug in the old code, where the program wont start with an empty / new database - this should be covered in a unit test Unit tests should be a separate issue
• Access to the database should be done without multiple database objects, if necessary there should be a singleton
• Make sure it is possible to store SM information with the vocable (i.e. degree of knowing) even if not used right now in any algorithm

Docker containers work well on github after some fiddling. So this seems to be a good way to build releases and maybe even crosscompile.

• Dockerfile for debian binary
• Dockerfile for android apk

• Document old algorithm
• Implement the algorithm used in old coleitra. This should get us started for a working version.
  There should be a real function interface so we can change between algorithms easily and on the fly. (Postponed for later milestone).

coleitra version

0.1

Operating System

Linux Debian Stable

Git commit

No response

Description

No response

Steps to reproduce

Lookup a word without flection section (i.e. olla for spanish)

Perceived severity by user

No response

Additional information

No response

Code of Conduct

• I agree to follow coleitras's Code of Conduct

It should be improved over the old coleitra regarding the following points:

• Text should be scaled to fill available space as large as possible (with necessary line breaks)
• Multiple meanings should be handled correctly (it should be possible therefore to divide space further and display multiple vocables in the divided spaces - a minimum size where a scrollarea is implemented should be postponed for a later version though)

Implement all stuff needed to just use json files to configure / extend different language support in grammarprovider

• Make reasonable data structure
• Load json files found into such list of data structures
• Implement calling function mechanism for this data structure