Inconsistency between mac and Windows output for SIL Hebrew

[Marc]

I’ve split this topic out from https://community.software.sil.org/t/issue-with-holem-vs-holem-for-waw-with-the-sil-hebrew-keyboard/5164 in order to separate the keyboard design discussion from a potential issue with Keyman.

This topic is about potential difference in behaviour between Windows and macOS. The original topic continues the keyboard design issue.

@martinacc wrote originally:

Hi,
I have an issue with Holem vs Holem for Waw with the SIL Hebrew keyboard.

Normally, typing “O” will output Holem, and typing “Shift + O” will output Holem for Waw.
I wish this mapping is consistent. However, the combination with Waw messes it up.
When I type “Waw” followed by “O”, it automatically turns into “Waw + Holem for Waw”.
I need to first type “O”, then “Waw” in order to output “Holem Waw”.

This is not how I normally type. I used to use the SBL Hebrew keyboard, and typing “O” always outputs “Holem”, even after “Waw”. I’ve always instructed my students to first time “Waw” then “Holem” for “Holem Waw”. Now, I will have to tell them NEVER to do it for “Holem Waw”. It is very important, because if students type “Holem for Waw”, their answers will be marked as incorrect by the online grading system.

Is it possible to prevent the automatic transformation?

Edit:
to make it worse, the auto transformation issue happens to the Mac version only. The windows version does not have this issue. So there is in consistency.

PS: I switched from the SBL Hebrew keyboard to Keyman, because the SBL Hebrew keyboard is inconsistent on the presentation of Shin and Sin. On Mac, the unicode value for Shin is \uFB2A and on Windows it is \u05E9\u05C1. Keyman is consistent on the presentation of Shin and Sin. But now, the “Holem” → “Holem for Waw” issue with the Mac version of Keyman is worse than the Sin/Shin issue with the SBL Hebrew keyboard in my work, I may have to painfully switch back to the SBL keyboard.

So, again, is there any way to prevent this automatic transformation of “Holem” to “Holem for Waw” after Waw on Mac?

Thanks!

… see original topic for more discussions …

Okay, I’ve tested some more this morning. I had a test version of Keyman installed on my mac last week, and it turns out I was seeing anomolous behaviour with that. Installing 14.0.280, I see the expected behaviour on macOS for sample key combinations do (gives U+05D3 U+05B9) and wo (gives U+05D5 U+05BA). This matches what I see for Windows.

So if you are seeing something different on Windows and macOS, we need to try and figure out why. Can we start off again with you trying those two sequences on both platforms and documenting what is being produced? Note the OS version, Keyman version and keyboard version in both cases. Use a basic app such as TextEdit on mac and Notepad on Windows, so we can be sure that there are no application interactions.

[martinacc]

Hi @Marc,

Thanks a lot for doing the test!
I did some tests again after reading your post. This time, I am seeing the consistent behavior on Win and MacOS.
I can confirm that do will produce U+05D3 U+05B9 and wo will produce U+05D5 U+05BA.

So, there is no problem, then, with Win and MacOS on this matter. The issue is with the mobile version now. I made the mobile version. Therefore, I know more about how it works. I mapped it according to the SBL Hebrew keyboard. o will consistently produce U+05B9 and ⇧ + o will consistently produce U+05BA.

Also, another behavior of the desktop Keyman keyboard is this: when we type ow, it will automatically switch to the sequence of U+05D5 U+05B9. The mobile version does not do it. It will keep the sequence: U+05B9 U+05D5.

I don’t know if there are other differences, but these two are that I’m aware of and certainly needs to be taken care of for the sake of consistency.

[martinacc]

Hi @Marc,

Can you test do wo?

From my understanding of the keyboard, we are expecting to output:
U+05D3 U+05B9 U+05D5 U+05BA. However, it outputs:
U+05D3 U+05B9 U+05D5 U+05B9

In this case, although the 2nd o immediately follows w, the keyboard does not convert it into U+05BA. This is problematic, because in this case, Waw (w) is preceded by a vowel. The o that follows it has to be U+05BA. This will be a bug, as I won’t be able to type U+05BA after the Waw in this case.

In this case, o for U+05B9 and ⇧ + o for U+05BA that the SBL keyboard features is better.

Edit: Just add a note: the combination of cowo, where c stands for any consonant, may not exist in the real world (I can’t think of one case), but I just want to bring this up to see if this is the intended output, because it clearly breaks the pattern.

[Marc]

@Lorna may be better placed to respond to keyboard design on this keyboard as she has been involved in the maintenance of the SIL Hebrew keyboard, whereas I haven’t – I’ve been working on Keyman itself.

Given the mobile keyboard follows the SBL Hebrew keyboard layout, it is probably better for it to be a separate keyboard with its own name (e.g. SBL Hebrew?) and then the desktop version of that keyboard can be tweaked to match your requirements and be consistent across all platforms.

[martinacc]

@Marc

Given the mobile keyboard follows the SBL Hebrew keyboard layout, it is probably better for it to be a separate keyboard with its own name (e.g. SBL Hebrew?) and then the desktop version of that keyboard can be tweaked to match your requirements and be consistent across all platforms.

Hi @Marc,

That’s a great idea.
If we can make both available, it’ll be great.

I made the SBL keyboard (SIL mapping) thinking it’s the same as the Keyman SIL Hebrew keyboard, because the SBL Hebrew keyboard says it is using the SIL mapping. I just discovered the Holem vs Holem for Waw two weeks ago… I had a Zoom meeting with students last night and this issue came up. For now, I made a user script to automatically convert any Holem for Waw students type into a regular Holem.

[Lorna]

I’m willing for us to do this.

• That would mean we would need to update the sil_hebrew keyboard to change the mobile layout to be consistent with the desktop.
• We would need to choose a new name for another Hebrew keyboard. We can’t use SBL because they have yet another layout. That keyboard can have the original SIL Hebrew keyboard desktop layout and the mobile that we have in sil_hebrew can be transferred to that new keyboard.

Any ideas on a good name? Hebrew (SIL) Original? LOL.

I can revert the sil_hebrew mobile layout. Would you want to modify the sil_hebrew to the holem behavior you want @martinacc? Or do you want me to do it and let you test before it is submitted? I’m okay either way, but it might be faster for you to get what you need if you feel comfortable doing the work.

[martinacc]

Hi Lorna,

Thank you very much.

Like you said in this post:

@Lorna

There was an intentional change made to the output of holem and holem with waw in 2007. It’s probable that the SBL keyboard (which was originally created on the SIL keyboard layout) was created prior to this change.

The latest version of the SBL Hebrew keyboard was last updated 4/21/2008.

Naming is tricky here because the SBL Hebrew keyboard calls it SIL layout too. So, there are the Keyman SIL Hebrew keyboard and the SBL SIL Hebrew keyboard.

@Lorna

I can revert the sil_hebrew mobile layout. Would you want to modify the sil_hebrew to the holem behavior you want @martinacc? Or do you want me to do it and let you test before it is submitted? I’m okay either way, but it might be faster for you to get what you need if you feel comfortable doing the work.

It was my first time to create a keyboard and it’s been a while since then. If you can take over, it’ll be great. I’ll be happy to test it.
Thank you very much!