This commit fixes an issue reported by @Sera4000 regarding duplicate logs. The problem was with the way the currently read offset was tracked. In the case when the blob had grown since the last time we read it, it happened that the size was set to 0 and not updated after reading, leading to the registry being saved with offset = 0 for the file. On the next iteration, the file would be read in full.
This PR makes it so that in each case (full read / partial read / empty file / error) we track the size of the chunk read in a variable delta_size and add that to the current file[:offset] to get the new offset.
This commit fixes an issue reported by @Sera4000 regarding duplicate logs. The problem was with the way the currently read offset was tracked. In the case when the blob had grown since the last time we read it, it happened that the
sizewas set to 0 and not updated after reading, leading to the registry being saved withoffset = 0for the file. On the next iteration, the file would be read in full.This PR makes it so that in each case (full read / partial read / empty file / error) we track the size of the chunk read in a variable
delta_sizeand add that to the currentfile[:offset]to get the new offset.