User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:188.8.131.52) Gecko/20071127 Firefox/184.108.40.206 Build Identifier: It would be very helpful if a user could modify what the spellchecker does. Perhaps there is an editable control file somewhere, if so, pls tell me where. Tks. What I would like to do is deal with the following: 1. Tell it better substitutions - why does it pony up yew instead of you, given the frequency of both words in English, for a common mistyping of you. Or it produces Shae (what the heck is that) for a mistyping of have. 2. Tell it about ; which is often (by me anyway) mistyped for '. Currently it handles don;t etc badly due to this, requires adjusting position, etc. 3. Make it pay attention to 1 and 2 letter words. If it doesn't know what to substitute for them, at least it should bring them up so the user doesn't miss them. Reproducible: Always Steps to Reproduce: 1. 2. 3.
Németh: where should these kind of bugs go? Would it be a bug for mozilla or hunspell?
There are problems with both of Mozilla and Hunspell. 1. Mozilla limits suggestions to 5, and it is a problem especially for the short misspelled words with more than 5 suggestions. Sorting by frequency would be an optimization here. 2. Need a sentence-level (or grammar) checker. 3. 2-letter words have been checked by Mozilla. 1-letter words (letter names) are frequent in monograms (in capitalized form) or in mathematical formulae. I believe, grammar checking would be better than forbid 1-letter words by word-level spell checker. Fortunately, Hunspell gives better suggestions for a lot of mistypes. These bugs can stay here or go http://hunspell.sourceforge.net.
I'm not sure what you're saying in #3. I don't think most people when sending email are including mathematical formulas, and are not monograms (IBM) devoid of whitespace internally? If the spelling corrector let thru a, i, and the beginnings of lists ( a. or (a) etc.), but flagged other single letters, I think that would be better than it's current functioning. Or it could detect if an area had +, - etc. in it and skip that as a formula. Or have a setting to do or not do this A couple of things the spelling corrector does that I forgot to mention: 1. It loses its place, so after some corrections it is looking at the middle and end of the next word. I am sort of sure that this is a bug that only started happening in the last few versions. 2. It thinks some "words" are okay when they are not, either not reporting them or ponying them up as substitutions. I do not recall off the top of my head any specific example. For awhile I looked in my personal dictionary to see if I had inadvertently added those at some point, but rarely found them, so they must be in a master dictionary.
I'm sorry, I called initials as "monograms", like H and G in "H. G. Wells". 1-letter words are real words, see http://www.m-w.com/dictionary/r for "r" for example. You are right, letter names are most frequent in ordered lists. Recognizing lists or other word structures we need a grammar checker. Unfortunately, Mozilla has contained only a word-level spell checker with limited personal dictionary features yet. I believe, future forbidden word list feature of the personal dictionary will be able to solve your problem with 1-letter words. Now you need to delete these 1-letter words from en-US.dic file of your Mozilla distribution manually. > 1. It loses its place, so after some corrections it is looking at the middle > and end of the next word. I am sort of sure that this is a bug that only > started happening in the last few versions. It's a serious bug. It would be great to attach a test case. Thanks in advance. 2. It thinks some "words" are okay when they are not, either not reporting them or ponying them up as substitutions. Word-level spell checking has its limits, for example it doesn't recognize "false friends" and mistypes resulting (other) dictionary words. But send your examples, and we will try to fix them. Thanks for your bug report.
Here is an example of the spellchecker losing its place and starting in the middle of a word, okay the text makes no sense, I was just trying to reproduce it: Original: a mouse arent says When it brings up arent substitute are not Then it positions itself on ays and suggests says
Here's another position bug: Original: Rgln yesterday When it brings up Rgln substitute Reglan Then it positions itself at esterday and suggests yesterday.
Severity: enhancement → normal
This problem was confirmed in Bug 261930, please vote for it on its page.
Please reduce this to one issue, and file additional bugs, one for each issue.
Component: Mail Window Front End → Spelling checker
Product: Thunderbird → Core
QA Contact: front-end → spelling-checker
Status: UNCONFIRMED → RESOLVED
Last Resolved: 10 years ago
Resolution: --- → FIXED
afaik none of these have been addressed. With no comment or change of summary it seems you are withdrawing the entire bug, so INVALID is the appropriate resolution
Resolution: FIXED → INVALID
It's not actually invalid, it's still glaringly wrong, but that was the only way I could see to stop getting email about an old report that no one seems to have any interest in fixing, so I don't want to be bothered about it any more.
Resolution: INVALID → WONTFIX
You need to log in before you can comment on or make changes to this bug.