consumes more bandwidth

[webp-sh]

consumes more bandwidth

[BennyThink]

Upon each request, we're using head method to get response headers. We need to do it in the first place because we don't know if the image has been changed.

HEAD method is lightweight, because it only contains http headers. 209B in this example

For your recommendation, there's a paradox:

instead of requesting the original picture first, find the webp file according to etag.
we need to know etag, then we can find webp image, and in order to find etag we need to request to remote.

And think of one edge case:

1. we already have one cache webp image
2. remote side has update the image for one same URL.
3. we would then return old webp which is not update-to-date.

So in general, this is intended.

[thetrueself]

Although the head method does not return content, frequent establishment of http requests is also a resource consumption. Due to the very low frequency of image changes with the same name, it is recommended to add the cache time configuration (optional parameter) to the configuration file. During the cache time, if the webp has been generated, there is no need to request the source image address.

[BennyThink]

Your cache time is a fixed value, which is unstable - by saying unstable I mean how could you make sure that the time you set will always return the correct image? It may be true on most cases, but I'm not wiling to take the chance.

And further more, let's track back to local mode

STAT, err := os.Stat(RawImagePath)

ModifiedTime := STAT.ModTime().Unix()
// webpFilename: abc.jpg.png -> abc.jpg.png.1582558990.webp
WebpFilename := fmt.Sprintf("%s.%d.webp", ImgFilename, ModifiedTime)

We have a few lines of code to get the modification time of one image, to help us keep track of whether our webp was outdated.

On remote mode, we take advantages of etag and that makes the behavior consistent - and this is as stable as ModTime().

Possibly we make take advantages of standard http cache headers like Cache-Control: max-age=31536000 or Last-Modified. But again, etag is one kind of cache validation mechanism and we're using it.

ref: https://developer.mozilla.org/en-US/docs/Web/HTTP/Caching