You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The type of the --function-arg-placeholders command line argument was changed from bool to int, to allow the default to be an "unset" value of -1 which allows the value specified using the new ArgumentLists config option to take effect.
However, I didn't realize at the time that the the option parser for bool is more permissive than the one for int: in addition to 0 and 1, it accepts true and false (and some capitalized versions thereof).
That means there are users using a released version of clangd today who have opted to use e.g. --function-arg-placeholders=false in their clangd settings, for whom, as things currently stand, clangd will refuse to start when clangd 20 is released and they upgrade to it.
We should do something to avoid this breakage before clangd 20 is released. One idea that comes to mind is to make the option string valued so it can accept both the previously accepted values and -1.
The text was updated successfully, but these errors were encountered:
That means there are users using a released version of clangd today who have opted to use e.g. --function-arg-placeholders=false in their clangd settings
Here is an example of this I've seen in the wild (it's actually how I discovered this regression).
This is a regression from llvm/llvm-project@18ca7ad (cc @MK-Alias)
The type of the
--function-arg-placeholders
command line argument was changed frombool
toint
, to allow the default to be an "unset" value of -1 which allows the value specified using the newArgumentLists
config option to take effect.However, I didn't realize at the time that the the option parser for
bool
is more permissive than the one forint
: in addition to0
and1
, it acceptstrue
andfalse
(and some capitalized versions thereof).That means there are users using a released version of clangd today who have opted to use e.g.
--function-arg-placeholders=false
in their clangd settings, for whom, as things currently stand, clangd will refuse to start when clangd 20 is released and they upgrade to it.We should do something to avoid this breakage before clangd 20 is released. One idea that comes to mind is to make the option
string
valued so it can accept both the previously accepted values and-1
.The text was updated successfully, but these errors were encountered: