>> In essence, it boils down to webbrowserpersist expecting to do two passes, and
>> to download files before determining the filename to replace the URI with,
>
> Yes, that's the right way to go about it.
Do you mean to say that you think that when CSS is being fixed up, FixupURI should block and not return until the original URI has been downloaded? Wouldn't forcing all the downloads to be made in serial like that have an adverse impact on performance?
Also, by "tests," do you mean automated tests, or just manually-verifiable testcases? I have no idea how to automate testing of webbrowserpersist...
>> In essence, it boils down to webbrowserpersist expecting to do two passes, and
>> to download files before determining the filename to replace the URI with,
>
> Yes, that's the right way to go about it.
Do you mean to say that you think that when CSS is being fixed up, FixupURI should block and not return until the original URI has been downloaded? Wouldn't forcing all the downloads to be made in serial like that have an adverse impact on performance?
Also, by "tests," do you mean automated tests, or just manually-verifiable testcases? I have no idea how to automate testing of webbrowserpersi st...