AI OnAI Off
I would really recommend to reconsider "serving" a IO stream from the disk directly to repsonse for every request. You should consider some sort of caching layer in between. ImageResizer (not promotion - don't have any relations with that company) library could be one of those.
Are you able to use Assets subsystem? The only downside of Assets as far as I remeber is fact that most of the stuff needs be done in code, and UI is for exploring only :) They have pretty extensible provider configuration system which provides you with set of built-in providers which one of them is Amazon S3 if you are targeting high-scale site.
I´ve implemented a metafield of type File and would like to provide a link to uploaded files for a product.
For images (which got it´s own metafield ImageFile) you can easily access the image and get url via following code for example:
Entry.ItemAttributes.Images.Image[0].Url
But the issue with the file metafield is that it seems it´s stored in database instead of saved to disk. Following code gets me the byte array:
Entry.ItemAttributes.Files.File[0].FileContents where FileContents is a bytearray.
So my question is what is best practice for handling File metafields? My solution that im thinking of is to create a aspx page where i stream file to disk and then response.write. The aspx file will take EntryID and i.e an GUID as a unique identifier so my url will be like:
http://mypage.com/mypage.aspx?entryid=50&guid=324234
Is there a more convinient solution? Use generic handlers instead?