I want to add that I only tested this patch with the patch from the other bug also applied, though they can be test separately.
With both patches applied, I saw even better results.
Doing the srfsh script to request a single bib record with fleshed call numbers did not increase the drone size in any appreciable way.
The query from comment #2 on this bug only increased the drone size by a few hundred bytes at a time, suggesting that there is still a small memory leak somewhere. However, I think this patch makes a significant improvement, so I have pushed a signoff branch to user/dyrcona/lp1974195_fix_flesh_fm_search_leak-signoff:
I tested the branch on this bug using the methodology described in https:/ /bugs.launchpad .net/opensrf/ +bug/1974193/ comments/ 3.
I want to add that I only tested this patch with the patch from the other bug also applied, though they can be test separately.
With both patches applied, I saw even better results.
Doing the srfsh script to request a single bib record with fleshed call numbers did not increase the drone size in any appreciable way.
The query from comment #2 on this bug only increased the drone size by a few hundred bytes at a time, suggesting that there is still a small memory leak somewhere. However, I think this patch makes a significant improvement, so I have pushed a signoff branch to user/dyrcona/ lp1974195_ fix_flesh_ fm_search_ leak-signoff:
https:/ /git.evergreen- ils.org/ ?p=working/ Evergreen. git;a=shortlog; h=refs/ heads/user/ dyrcona/ lp1974195_ fix_flesh_ fm_search_ leak-signoff
NOTE: My signoff has been cherry-picked on top of current master as that is how I tested it.