On Thu, Jul 18, 2019 at 12:49 PM Ulrich Windl <Ulrich.Windl@xxxxxxxxxxxxxxxxxxxx> wrote:
>>> Mantas Mikulenas <grawity@xxxxxxxxx> schrieb am 18.07.2019 um 10:46 in
Nachricht
<CAPWNY8W1JKV6TeSSg_dD82+QfM3T2_o=KxpwiuZDVfbxx4Y-LA@xxxxxxxxxxxxxx>:
> On Thu, Jul 18, 2019 at 11:34 AM Ulrich Windl <
> Ulrich.Windl@xxxxxxxxxxxxxxxxxxxx> wrote:
>
>> Hi!
>>
>> I noticed that a line of "=======" in "[Service]" cases the message "
>> Unknown lvalue '' in section 'Service'".
>> (systemd 228)
>>
>> Shouldn't that be "Parse error at '=======' in section 'Service'"?
>>
>
> Arguably it isn't a parse error – the keyfile parser successfully
> recognizes the line as assigning the value "======" to the key "". It's
> only later when the parsed results are interpreted that each key is matched
> to an internal handler.
Well, of course it's a matter of interpretation:
You could assign "======" to a variable with an empty name, or you could
assing "=====" to a variable named "=", but to me the the parser sould detect
theis kind of error (I susüect it was a mark from some merge, btw)
>
> The error message *could* be clearer if all such errors had a common "Parse
> error:" prefix, I guess. (And what's the point of calling it an 'lvalue'
> anyway?...)
Well if the variable with an empty name is valid, it's an lvalue and it's name
is the "unknown" value (if you like to argue that way). IMHO variable names
consisting of empty strings are a bad concept at least...
Yes, and Zbigniew's PR adjusts the parser to not allow empty names.
Mantas Mikulėnas
_______________________________________________ systemd-devel mailing list systemd-devel@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/systemd-devel
