How much prose do you want in one line? The name can not be a number, not something that could be a cell reference in any of the possible reference styles, and it can't start with a digit, which here is the case.

Emil: please propose a smashing new one-liner per Eike's parameters.

Depending on the expected behavior:
Invalid name. Only use letters, numbers and underscore in this order. 
Or:
Invalid name. Only use letters, numbers and underscore, starting with a letter.

Repro 70+.

The dialog obviously applies too much restriction. E.g., it's possible to use names like "a,b" (i.e., with comma) - just type that into the Name Box above the cells; but the dialog has that odd restriction. IMO it should be fixed to match the actual restriction first, then the message should be fixed.

Additionally, you may start with e.g. underscore. So I suppose that you can't come with a single error message short enough. The dialog possibly could try to be smart enough to show different error messages based on the specific problem - like "The name can't start with a digit"; "The name entered is a cell reference"; ...

Roman Kuznetsov committed a patch related to this issue.
It has been pushed to "master":

https://git.libreoffice.org/core/commit/949d648594f441dbb7e9c600d7925fba7e882ff9

tdf#112279 Change the error message to more clear

It will be available in 7.1.0.

The patch should be included in the daily builds available at
https://dev-builds.libreoffice.org/daily/ in the next 24-48 hours. More
information about daily builds can be found at:
https://wiki.documentfoundation.org/Testing_Daily_Builds

Affected users are encouraged to test the fix and report feedback.

Roman, is this bug solved?

(In reply to Mike Kaganski from comment #7)
> Additionally, you may start with e.g. underscore.

FWIW, that is not what the error message currently says:

https://git.libreoffice.org/core/+/refs/heads/master/sc/inc/globstr.hrc

NC_("STR_ERR_NAME_INVALID", "Invalid name. Start with a letter, use only letters, numbers and underscore.")

How do you think about this?

NC_("STR_ERR_NAME_INVALID", "Invalid name. Start with a letter or an underscore, and use only letters, numbers and underscore.")

(In reply to BogdanB from comment #11)
> How do you think about this?
> 
> NC_("STR_ERR_NAME_INVALID", "Invalid name. Start with a letter or an
> underscore, and use only letters, numbers and underscore.")

The problem is that the name cannot start with a number nor be equivalent to an address (in any form) either, and the message has to be short.

Starting always with an underscore would solve these naming conflicts, but Calc should not impose that.

While not completely imposing the underscore as first character, the short message could still hint such naming method as a simple solution for those cases that trigger some naming conflict.

I do not know whether the possible conflict with cell address (A1 or R1C1 or R[1]C[1], although brackets are not allowed) is also expected to be handled by the same message, or there is a separate one for that instead.


So, some possibilities I write here for discussion (attention to the order of terms):


1. Use only underscores, digits and letters. Do not start with digit. Cell address not allowed.

2. Use only underscores, digits and letters. Do not start with digit. Avoid cell address.

3. Use only underscores, digits and letters. Do not start with digit.

4. Use only "_", digits and letters. Do not start with digit.

5. Use only "_", [1-9] and letters. Do not start with digit.

6. Use only "_", [1-9] and letters. Do not start with digit. Avoid cell address.

7. Use only "_", [1-9] and letters. Do not start with digit. Cell address not allowed.



The first alternative covers more and is more clear, but longer (92 chars).

The third alternative is shorter (66 chars), in case the conflict with cell addresses is handled by a separate message – I don't know the answer to that.

Others are compromises in some way or another, with different lengths.

I like "3. Use only underscores, digits and letters. Do not start with digit." but with numbers instead of digits: "3. Use only underscores, numbers and letters. Do not start with numbers."

(In reply to BogdanB from comment #13)
> I like "3. Use only underscores, digits and letters. Do not start with
> digit." but with numbers instead of digits: "3. Use only underscores,
> numbers and letters. Do not start with numbers."

"Digit" is shorter than "number", and plural is not needed (nor wanted) at the end of that sentence. That's why :). I don't know what's the acceptable length for the (English) message.

The point about cell addresses is also relevant. Is that part handled by some other message? If it isn't, and if the length is not a problem for alternative 1, I would go with it:

Use only underscores, digits and letters. Do not start with digit. Cell address not allowed.

If the cell address point is not handled by some other message and the length is a problem for alternative 1, then alternative 7:

Use only "_", [1-9] and letters. Do not start with digit. Cell address not allowed.

or:

Use only "_", 1...9 and letters. Do not start with digit. Cell address not allowed.

It all depends on the restrictions, which I don't know.

(In reply to BogdanB from comment #13)
> numbers instead of digits

Please do not confuse "numbers" (anything like e.g. "123" or "0.15*10^-123") with "digits" (= symbols used do represent the math decimal notation elements). Indeed, the "decimal" is implied, and we also do not consider e.g. Chinese 一二三.

(In reply to Mike Kaganski from comment #16)
> "12f" is gladly accepted when assigned using
> the Name Box of main UI.

It is rejected for me.

Version: 7.6.0.0.alpha0+ (X86_64) / LibreOffice Community
Build ID: daf30c29be67b8b8fa361b0efd1a6cdbe087b6f8
CPU threads: 4; OS: Windows 10.0 Build 19044; UI render: Skia/Raster; VCL: win
Locale: en-US (es_AR); UI: en-US
Calc: CL threaded