Timeline for Write a compact spelling checker
Current License: CC BY-SA 3.0
31 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Nov 22, 2015 at 2:29 | answer | added | wolfhammer | timeline score: 1 | |
| Nov 21, 2015 at 12:55 | history | edited | user45941 |
edited tags
|
|
| Nov 21, 2015 at 7:47 | answer | added | Adam Katz | timeline score: 0 | |
| Nov 21, 2015 at 5:25 | comment | added | Adam Katz |
Pastebin uses DOS line breaks, so the file they host is 6937 bytes. You can recreate the 5942 size by converting it. (MD5 of the 5942 file is 673911d27b213a0812598bf3b6b6c9ee)
|
|
| Jun 14, 2015 at 18:05 | history | tweeted | twitter.com/#!/StackCodeGolf/status/610145985059233792 | ||
| Jun 7, 2015 at 5:36 | answer | added | Reto Koradi | timeline score: 1 | |
| Jun 6, 2015 at 4:59 | answer | added | Gaslight Deceive Subvert | timeline score: 1 | |
| Jun 5, 2015 at 15:14 | history | edited | CJ Dennis | CC BY-SA 3.0 |
Some people need everything spelt out for them...
|
| Jun 5, 2015 at 14:16 | answer | added | CJ Dennis | timeline score: 2 | |
| Jun 5, 2015 at 13:44 | comment | added | CJ Dennis | If it's obvious that more than 50 words will break the code they don't need to be specified individually. It's only if someone has a fuzzy solution and is running close to the wind. A few extra words might be enough to break their code. | |
| Jun 5, 2015 at 13:40 | comment | added | r3mainer | @CJDennis That sounds more like cops-and-robbers than code-golf | |
| Jun 5, 2015 at 13:37 | comment | added | Reto Koradi | True. But then they can update their code again to exclude those. So in the end, you will still have 1000 words. Also, I have a feeling that if anybody systematically searches for false positives for a given solution, any solution that has false positives will have at least 50 of them. I doubt that solutions with false positives are viable, at least as long as somebody puts an effort into finding enough words that break them. | |
| Jun 5, 2015 at 13:30 | comment | added | CJ Dennis | @RetoKoradi Anyone can invalidate your code by proving more than 50 false positives so I don't think this is a concern. | |
| Jun 5, 2015 at 13:26 | comment | added | Reto Koradi | I think the list of tests for false positives will need to be at least as large as the list of valid words. Otherwise it will be easier/shorter to encode the list of invalid words. | |
| Jun 5, 2015 at 13:24 | comment | added | r3mainer | Re: If a garbage string doesn't produce a false positive on anyone's code (or it only produces a false positive on an entry that has too many false positives anyway) that word won't be added to the list. — this approach is really going to put people off submitting entries before a list appears. As things stand, my answer is the only one that will generate false positives, so your list will consist exclusively of words that my code can't handle. Seems a bit unfair... | |
| Jun 5, 2015 at 12:39 | answer | added | grc | timeline score: 1 | |
| Jun 5, 2015 at 10:53 | comment | added | CJ Dennis | @trichoplax I know! I specifically checked for duplicates before writing this question and that question was posted while I was writing this one! | |
| Jun 5, 2015 at 10:48 | comment | added | trichoplax is on Codidact now | Not a duplicate, but what a weird coincidence | |
| Jun 5, 2015 at 10:35 | history | edited | CJ Dennis | CC BY-SA 3.0 |
Fewer words in list!
|
| Jun 5, 2015 at 10:31 | comment | added | CJ Dennis | @squeamishossifrage Well done for spotting that! I'll update! | |
| Jun 5, 2015 at 10:28 | comment | added | r3mainer |
The list only contains 997 unique words — help and smoke both appear twice
|
|
| Jun 5, 2015 at 10:21 | answer | added | Nejc | timeline score: 3 | |
| Jun 5, 2015 at 9:33 | history | edited | CJ Dennis | CC BY-SA 3.0 |
Explained capitalisation
|
| Jun 5, 2015 at 9:19 | comment | added | CJ Dennis |
@Sp3000 Would you accept that as correctly spelt in a book or an email? I'll go as far as COLOUR which is optional. That is to say COLOUR may produce TRUE or FALSE depending on the intent of the programmer. COLOR must always produce FALSE. Any other weird capitalisation must produce FALSE as well. So, colour and Colour must be TRUE, COLOUR may be either TRUE or FALSE (but consistently across all words) and color, Color, COLOR, CoLoUr etc. must always produce FALSE.
|
|
| Jun 5, 2015 at 8:27 | comment | added | Sp3000 |
If Colour returns TRUE, what about cOlOUr?
|
|
| Jun 5, 2015 at 8:09 | history | edited | CJ Dennis | CC BY-SA 3.0 |
Added error list condition
|
| Jun 5, 2015 at 8:05 | comment | added | CJ Dennis |
I will compile a list of specific, known false positives as the challenge progresses, i.e. "checking the word XXX break's Fred's code". Since this is supposed to be a spelling checker and there's nothing to stop real people from typing garbage the program should accept all garbage words and output FALSE. If a garbage string doesn't produce a false positive on anyone's code (or it only produces a false positive on an entry that has too many false positives anyway) that word won't be added to the list.
|
|
| Jun 5, 2015 at 7:42 | comment | added | isaacg |
Sorry, I wasn't clear. I meant what if daskjhaasudftaigusdfasdfasdasuhdfasdfasdfasdad happens to be accepted by the program - is that a false positive? Or is there some specific, finite list of words we can check to count false positives?
|
|
| Jun 5, 2015 at 7:37 | comment | added | CJ Dennis | @isaacg Assume the input is always a single word, all ASCII [A-Za-z]. I'm pretty sure there are only single words in the list, no non-alphabetic characters. | |
| Jun 5, 2015 at 7:35 | comment | added | isaacg | What is the space where false positives can come from? All strings? It's hard to count false positives if you don't know what to check. | |
| Jun 5, 2015 at 7:30 | history | asked | CJ Dennis | CC BY-SA 3.0 |