STR: * open SMS app * click on "+" to create a new SMS * try to type "bla ?" (with a space before '?') Expected: * "bla ?" is written Actual: * "bla? " is written Not sure if this should be a blocker, but this is surely annoying.

marking qawanted because this might happen elsewhere and/or with other characters. I verified this happens with '!' as well.

Tested with BUILD 20130108070203 It happens in composing also. I believe it's a "force" auto-correction. The punctuation includes "!", ",", ".", "?"

Hi, all, Thanks for your help! I cannot reproduce this bug on the latest v1.1.0 hd build. * Test Build:(Mozilla-b2g18_v1_1_0_hd-unagi/2013-07-01-07-02-15) + Mercurial-Information: - Gecko: "f686f56c11d6" - Gaia: "" + Git-Information: - Gecko: "" - Gaia: "c7472acec84f0d4527cdd6fd555d289e1d3e1d1d"

Unfortunately I see it back on master now :( Not sure if that happens on v1.1 though.

I think this sort of rules should be layout-dependent.

Hi, Julien, Sorry! I misunderstood this case. Yes, we need to reopen it since the bug still exists. I also can reproduce it on v1-train. I think it is unlike bug 827414. Please don't mark as a duplicate of bug 827414. Thanks!

Hello folks. I'm teaching a person how to solve bugs in Gaia, can you let me this as an example? Thanks you.

Assigning to :mai. Now she will explain her thoughts about the issue. :mai just recently joined the FE team here in TID. So say welcome folks! Thank you.

Hi all, I think the auto-corrector works correctly when it makes the corrections with punctuation chars. For me, the problem is when you try to revert the punctuation corrections, because the punctuation char is removed. For example: The user writes: "bla ?" The autocorrector converts the text to "bla?" The user press the key "Backspace": * The expected output: "bla ?" * The actual output: "bla " WDYT?

I think the same here. Anyway, maybe the guys from UX can share their opinion...

(In reply to marina rodríguez [:mai] from comment #10) > Hi all, > I think the auto-corrector works correctly when it makes the corrections > with punctuation chars. For me, the problem is when you try to revert the > punctuation corrections, because the punctuation char is removed. > For example: > > The user writes: "bla ?" > The autocorrector converts the text to "bla?" > The user press the key "Backspace": > * The expected output: "bla ?" > * The actual output: "bla " > > WDYT? That's not the issue for me. It shouldn't autocorrect "bla ?" to "bla?" in the first place. It might make sense in English or Spanish but not in some other languages (French for example).

Actually that rule should be language-dependent.

Yes Julien, you're right. Anyway, current behavior is inconsistent with the expected behavior for cancelling an autocorrection action. After an autocorrection happens, if the user press backspace it should be returned to the state just before the correction. This is not the case here and it does not allow the user to type certain character sequences. I think we should file another bug to address this specific problem.

I filed the bug 905946 to address the specific problem of the inconsistent behavior with the punctuation chars autocorrection.

(In reply to Anthony Ricaud (:rik) from comment #13) > That's not the issue for me. It shouldn't autocorrect "bla ?" to "bla?" in > the first place. It might make sense in English or Spanish but not in some > other languages (French for example). The reason it does that is that when you tap on a word suggestion, the word is inserted with a space following it. So if you want punctuation after that word, it needs this automatic fix. It also happens when the user manually types a space followed by a question mark, but that is really a side-effect. It is not obvious to me how we are going to implement language-dependent punctuation rules in a general manner, but if the latin keyboard is violating language conventions, we ought to have a bug about it where we can collect a list of the desired behavior for each language.

David, you mean a different bug than this one ? Like one bug to do the general code with eg the french and american rules, and one bug to put the other language conventions ?

Is there any launch country for 1.4 that will have different punctuation rules?

We're planning this in bug 956180 by the way.

Should we dupe one or the other then?

It depends, bug 956180 is a bug for the new keyboard. It might not land in 1.4. If we need it in 1.4 we might need to backport it to the current one. So, yeah I don't know :p

Ok, actually, I don't mind whether it's 1.4 or 1.5, but I'd like to see some progress ;)

Backlogging so it isn't lost. If the refactor fixes this we can revisit.

Julien, is this a dup of bug 808740?

Yes it could be, I'd need to check on a new build to see if this is resolved. In the mean time, let's dupe and I'll reopen if I still see the issue.