Hi, I'm currently graphing something with dot, and find convenient to use the “autoreload” feature of ristretto. I open once the generated svg with ristretto, then edit my dot file, generate the svg, and automagically ristretto reloads it and display the result. But doing this, I discovered a (quite huge) memory leak, I guess somewhere in the reload code (or maybe in the closing or opening, not sure what is run during reload). I can reproduce the problem with: corsac@miria: import -w root trouc.png; ristretto trouc.png & sleep 1; echo "Before:" ;grep "Vm(Size|RSS|Data)" /proc/$(pidof ristretto)/status;import -w root trouc.png;sleep 1;echo "After:"; grep "Vm(Size|RSS|Data)" /proc/$(pidof ristretto)/status;import -w root trouc.png;sleep 1;echo "After:"; grep "Vm(Size|RSS|Data)" /proc/$(pidof ristretto)/status [1] 7581 Before: VmSize: 158600 kB VmRSS: 22756 kB VmData: 13676 kB After: VmSize: 162440 kB VmRSS: 26596 kB VmData: 17516 kB After: VmSize: 173960 kB VmRSS: 38116 kB VmData: 29036 kB Nothing is freed before/after a reload, so in my case it quickly lead to a huge occupation. I didn't understand why the box was that slow with 2GB ram, until OOM killed ristretto :) Cheers,
Created attachment 1916 valgrind log G_DEBUG=gc-friendly G_SLICE=always-malloc valgrind --leak-check=full --log-file=ristretto.log -v ristretto trouc.png and then some import -w root trouc.png to force reload.
I think I fixed it in revision 5744. Could you confirm?
(In reply to comment #2) > I think I fixed it in revision 5744. Could you confirm? Looks fixed, thanks!
ok, closing this bug.