When browsing a page with several images on a low memory machine, it may happen that the decompressed sizes of all images exceed the available memory, causing major swapping of pages which would make Dillo and the whole system slow. In that case, the user should be able to indicate a limit that should not be exceeded by default to load images (when load_images is set to YES).

Similarly, we have an ever increasing cache for page resources (HTML, compressed images, CSS, ...) which is never freed until the browser is closed. This behavior is generally convenient when the network is expensive but the RAM is not. In the opposite scenario, we want to remove elements from the cache after a given limit so we keep the cache under control.

By far, the dicache is the largest problem, as decompressed images take much more memory than any other resource, even after hours of browsing.

To implement a dicache limit, we first need to identify at which point we have enough information to stop further images from decompressing. When we are parsing <img> tags from the HTML page, there is no information yet on the image dimensions, so we cannot infer which size they will need after decompression.

Worse, it may happen that we are in the middle of the decoding of several images and we pass the image cache limit. At that point, we either discard the decoded image so far or we continue the decoding exceeding the limit. I would be inclined to go the safe route and abort any current decoding as soon as we reach the cache limit. At that point, only images that are requested by clicking with the mouse are decoded.

We could use a_Dicache_write().