about summary refs log tree commit diff stats
path: root/Manual_tests.md
blob: 068552150df98a241f5ec7fb71d2080c3c999b9e (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
pre { line-height: 125%; }
td.linenos .normal { color: inherit; background-color: transparent; padding-left: 5px; padding-right: 5px; }
span.linenos { color: inherit; background-color: transparent; padding-left: 5px; padding-right: 5px; }
td.linenos .special { color: #000000; background-color: #ffffc0; padding-left: 5px; padding-right: 5px; }
span.linenos.special { color: #000000; background-color: #ffffc0; padding-left: 5px; padding-right: 5px; }
.highlight .hll { background-color: #ffffcc }
.highlight .c { color: #888888 } /* Comment */
.highlight .err { color: #a61717; background-color: #e3d2d2 } /* Error */
.highlight .k { color: #008800; font-weight: bold } /* Keyword */
.highlight .ch { color: #888888 } /* Comment.Hashbang */
.highlight .cm { color: #888888 } /* Comment.Multiline */
.highlight .cp { color: #cc0000; font-weight: bold } /* Comment.Preproc */
.highlight .cpf { color: 
I care a lot about being able to automatically check _any_ property about my
program before it ever runs. However, some things don't have tests yet, either
because I don't know how to test them or because I've been lazy. I'll at least
record those here.

Initializing settings:
  - delete app settings, start with a filename; window opens running the text editor with cursor at top of file
  - run with absolute file path; quit; restart; window opens running the text editor in same position+dimensions
  - run with relative file path; quit; switch to new directory; restart without a filename; window opens running the text editor in same file in same position+dimensions
  - run with a filename on commandline, scroll around, quit; restart without a filename; window opens running the text editor in same position+dimensions
  - run with a filename on commandline, scroll around, quit; restart with same filename; window opens running the text editor in same position+dimensions
  - run with a filename on commandline, scroll around, quit; restart with new filename; window opens new filename with cursor up top
  - run editor, scroll around, move cursor to end of some line, quit; restart with new filename; window opens running the text editor in same position+dimensions
  - quit while running the text editor, restart; window opens running the text editor in same position+dimensions
  - quit while editing source (color; no drawings; no selection), restart; window opens editing source in same position+dimensions
  - start out running the text editor, move window, press ctrl+e twice; window is running text editor in same position+dimensions
  - start out editing source, move window, press ctrl+e twice; window is editing source in same position+dimensions
  - no log file; switching to source works

Code loading:
* run love with directory; text editor runs
* run love with zip file; text editor runs

* How the screen looks. Our tests use a level of indirection to check text and
  graphics printed to screen, but not the precise pixels they translate to.
    - where exactly the cursor is drawn to highlight a given character
    - analogously, how a shape precisely looks as you draw it

* start out running the text editor, press ctrl+e to edit source, make a change to the source, press ctrl+e twice to return to the source editor; the change should be preserved.

### Other compromises

Lua is dynamically typed. Tests can't patch over lack of type-checking.

* All strings are UTF-8. Bytes within them are not characters. I try to label
  byte offsets with the suffix `_offset`, and character positions as `_pos`.
  For example, `string.sub` should never use a `_pos`, only an `_offset`.

* Some ADT/interface support would be helpful in keeping per-line state in
  sync. Any change to line data should clear line `fragments` and
  `screen_line_starting_pos`.

* Some inputs get processed in love.textinput and some in love.keypressed.
  Several bugs have arisen due to destructive interference between the two for
  some key chord. I wish I could guarantee that the two sets are disjoint. But
  perhaps I'm not thinking about this right.

* Like any high-level language, it's easy to accidentally alias two non-scalar
  variables. I wish there was a way to require copy when assigning.

* I wish I could require pixel coordinates to be integers. The editor
  defensively converts input margins to integers.

* My test harness automatically runs `test_*` methods -- but only at the
  top-level. I wish there was a way to raise warnings if someone defines such
  a function inside a dict somewhere.