I kind of don't like the more complicated implementation for txt2img_upscale and I don't understand why it's needed.

The change from script classname to script name is fine but I'd rather it was a single PR.

And as for addition of two other functions for settings script args, I'm not sure if they are needed too.

I kind of don't like the more complicated implementation for txt2img_upscale and I don't understand why it's needed.

this I believe is implemented what you initially tried to implement then accidentally broke
irrc you initially try to set the seat using the set_named_arg you added for that specific purpose
but then you implemented something else which means the setup is already executed
so I implemented stuff to do every argument setting before it even creates p

the main point of this is that this makes setup() behave exactly the same as normal txt2img otherwise
there will be slight differences that might be important for certain extensions, yes those extension most likely can adapt but
it can also cause some potential bugs that is not going to be caught after a long time

I see this as slightly complicated it's here but reduce complications elsewhere

And as for addition of two other functions for settings script args, I'm not sure if they are needed too.

update_script_args it's just more convenient if you wash update the entire script arg
where as set_named_arg can do only do one-by-one

if you only need to update a small portion then short one by one is fine but it's totally possible that someone would want to update the entire arg for whatever reason
I think it's generally useful function for extensions / scripting

and get_script_args is is just the counterpart to set args, they might be some cases that we should just get the arcs but not set them, if I said a script needs to know what other script args are then this will come in handy

if by overcomplicating you mean that updating the entire seeds arg

I could agree with that the other orgs such as resize and variable seed strength doesn't need to be updated from the infotext
the only real two important values are the seed and sub-seed
but since I was doing that I decided to just implement it all for the sake of OCD
this can be changed to only update the seed and sub-seed easily

tbh I also have a bit of doubt that I don't know if I should update the entire seed script

The change from script classname to script name is fine but I'd rather it was a single PR.

I'll make it into a different po if this one is completely rejected

And as for addition of two other functions for settings script args, I'm not sure if they are needed too.

theres also an issue with set_named_arg
the script arg you want to set has to have a pre-defined arg_elem_id

update_script_args is more universal in that it will work on all scripts

this I believe is implemented what you initially tried to implement then accidentally broke irrc you initially try to set the seat using the set_named_arg you added for that specific purpose but then you implemented something else which means the setup is already executed so I implemented stuff to do every argument setting before it even creates p

the main point of this is that this makes setup() behave exactly the same as normal txt2img otherwise there will be slight differences that might be important for certain extensions, yes those extension most likely can adapt but it can also cause some potential bugs that is not going to be caught after a long time

I see this as slightly complicated it's here but reduce complications elsewhere

Well. We create p. It works just as usual. Setup runs just as usual. It sets the seed. Later in the program we change this seed. This is not abnormal. Custom scripts that create their own p will also do this (like XYZ plot). Why the need to touch the arguments set to p?

Well. We create p. It works just as usual. Setup runs just as usual. It sets the seed. Later in the program we change this seed. This is not abnormal. Custom scripts that create their own p will also do this (like XYZ plot). Why the need to touch the arguments set to p?

it could be just me thinking too much
even though I don't know if they're even is a use case maybe there's this weird extension but somehow absolutely relies on seed the properties being set by the UI and not change afterwards

this use case seems more ridiculously what I think about it

do you want me to revert setting to the way it was previously
if reverted I still think update_script_args get_script_args are still useful functions to keep around

For txt2img_upscale, yes, unless there's a reason to do this your way, I'd like to keep it simpler.

update_script_args get_script_args: I kind of don't want to do positional script args. If those functions works with a dictionary of name/value pairs, that would be nice.

The change from script classname to script name is fine but I'd rather it was a single PR.

moved to

• rework set_named_arg #14773

• remove everything else that is not strictly necessary ,leaving only force_enable_hr

For txt2img_upscale, yes, unless there's a reason to do this your way, I'd like to keep it simpler.
update_script_args get_script_args: I kind of don't want to do positional script args. If those functions works with a dictionary of name/value pairs, that would be nice.

removed for now