Re: [pacman-dev] [PATCH 4/4] Add download size to target list.

16 Oct 2007

      On 10/16/07, Dan McGee  wrote:
...
On 10/16/07, Nathan Jones  wrote:
...
On Mon, Oct 15, 2007 at 09:59:07PM -0500, Dan McGee wrote:
...
...
+unsigned long SYMEXPORT alpm_pkg_download_size(pmpkg_t *newpkg, pmdb_t *db_local)
+{
+       char *fpath = _alpm_filecache_find(alpm_pkg_get_filename(newpkg));
+       unsigned long size = 0;
+
+       if(fpath) {
+               size = 0;
+       } else if(handle->usedelta) {
+               alpm_list_t *deltas = pkg_upgrade_delta_path(newpkg, db_local);
+
+               if(deltas) {
+                       size = _alpm_delta_path_size_uncached(deltas);
Does this account for the delta is greater than 0.7 of the size case?
When finding the shortest delta path, the code does not look to the
cache to see if the file already exists. This function takes whatever
delta files will actually be used and calculates the remaining
downloaded size for the client to display, using 0 if a delta exists in
the cache.
...
...
-               /* round up if size is really small */
-               if(mbrsize < 0.1) {
-                       mbrsize = 0.1;
-               }
Any reason for removing this?
I removed it because it doesn't make as much sense with the download
size. For example, downloading a 30K package would display 100K for the
total package size but 30K for the download size.
The download size could be rounded as well, but it is possible for it to
be 0K if the file already exists in the cache, so displaying 100K would
not be correct.
I was thinking the installed size/package size always displayed as MB.
I know that when downloading we use K until we reach a certain point,
but I thought the initial display of sizes is done all in MB.
-Dan
Some example output:

$ pacS 855resolution
resolving dependencies... done.
looking for inter-conflicts... done.

Targets: 855resolution-0.4-5

Total Package Size:   0.10 MB

Proceed with installation? [Y/n]

Since we print the values to 2 decimal places, we should probably
round to 0.01 instead of 0.1. Or like you said, we could just get rid
of the rounding. I'll do this on my local branch, I like your original
thinking. :)

-Dan