Calling version of already-uploaded file doesn't store the conversion

[xuloo]

I have an uploader with this version:

version :png, if: :gif? do
  process convert: 'png'
end

(the gif? method reliably identifies a gif (this isn't the problem).)

the application has recently stopped users uploading gifs - so we're having to deal with gifs that were uploaded before the ban when creating a PDF with prawn. So now, when requesting an image we check to see if the original image was a gif, if so we redirect them to the 'png' version.

So far so good. Only the png version isn't working as expected - we're storing image on S3 and for an original image url

https://amazon-blah.com/bucket/image.gif 

we get given back the url

https://amazon-blah.com/bucket/image_png.gif 

and the image isn't stored to S3

I did a bit of googling and changed the png version to

version :png, if: :gif? do
  process convert: 'png'
  def full_filename (for_file = model.source.file)
    super.chomp(File.extname(super)) + '.png'
  end
end

which now gives me the url

https://amazon-blah.com/bucket/image_png.png

but, again the image isn't stored - so it can't be retrieved from the url so i get a 403 response from S3.

I've also tried catching that 403 at the point in the code where the PDF generator is trying to open the image url and calling

recreate_versions!(:png)

in the hope that would trigger the generated image to be stored - but no luck there, either. It can't be a permissions problem with S3 because the original images are uploaded just fine.

How do i get this version to be created AND stored? Am i missing a config setting? Is there something else daft i'm missing? Thanks

[thomasfedb]

Any ideas @bensie?

[rsiddle]

I am running into a similar problem. When processing media, the uploaded file is always the original filename with extension, however the file does have the correct underlying format. It appears to be just a naming issue?

[mshibuya]

#2659 should have solved this.