updated BUGs in general to remove older / fixed BUGs relating to the confusion of current/eids, etc.

update amendments in tables.sql to include job_id in entry_ammendment
added amend.py to move amendment-related code into its own file when we create a job (NewJob)
  and that job matches an amendmentType (via job_name or job_name:amt <- where amt relates to how we do a transform_image), then
  we enter a new EntryAmendment pa_job_mgr knows when a Transform job ends, and removes relevant EntryAmendment
files*.js use EntryAmendment data to render thumbnails with relevant AmendmentType
if a normal page load (like /files_ip), and there is an EntryAmendment, mark up the thumb, run the check jobs to look for completion of the job,
  removeal of the EntryAmendment and update the entry based on 'transformed' image

OVERALL: this is a functioning version that uses EntryAmendments and can handle loading a new page with outstanding amendments
  and 'deals' with it.  This is a good base, but does not cater for remove_files or move_files
This commit is contained in:
2025-10-20 19:31:57 +11:00
parent 905910ecf0
commit 56771308a6
8 changed files with 203 additions and 132 deletions

19
BUGs
View File

@@ -1,15 +1,7 @@
### Next: 141
### Next: 143
BUG-142: after transforming, the face data is still in the old spots, really should delete it / make it recalc
BUG-141: can currently try to flip a video (in a highlighted group)
BUG-140: When db is restarted underneath PA, it crashes job mgr... It should just accept timeouts, and keep trying to reconnect every 2? mins
BUG-139: using any large entry list and going next a few times, ends say 4 pages of 50 into 4000 matches (entries from DB < 50)...
- confirmed this is when person has 2 or more refimgs:
- on page "2", we get 49 pulled back in the ORM instead of the 50 expected -- b/c I use that to indicate we must be at the end of the list if not 50 found
-- really, need to fix once and for all the eids / re-running query.
do GetEntries as we do now, once done however, get all entry ids. Stick those into the DB with a unique query-id and datestamp
new func to get all details needed for entries in an eid list (of 1-page) - show this page of entries
use current, full eidlist and to work our start/end of list (next/prev), disabling.
then client can keep current page of data, if you hit next/prev, use DB unique query id / full list and page of eids, and give full data for new page of entries
Implications though, are if a search is invalidated (maybe delete / move a photo), need to remove them from the list on the DB too OR let user know/decide to fix/wait.
BUG-100: I managed to get 2 photos matching mich in the NOT_WORKING photo (probably dif refimgs but same p.tag?)
= /photos/2012/20120414-damien/IMG_8467.JPG
@@ -31,11 +23,6 @@ BUG-125: when an image is highlighted, then post the contextmenu on a different
There is a chance we need to change the document on click to a mouse down (or whatever the context menu
uses for default), rather than just fix the highlight
BUG-130: moving files and then trying to go next page and it got confused...
BUG-132: right arrow to go to next photo in viewer ALSO scrolls to the right, needs a return somewhere in the jscript
BUG-133: when rebuilding pa[dev], the first run fails to have symlinks to the right paths for Import/Storage, etc. a simple restart fixes - so potentially the intial copy or some other race condition?
BUG-134: when moving set of photos on page, then move another set of photos on page, the first set reappears. Could really delete them from the dom?
BUG-135: failed to rotate: 2006/20061215-ITS-xmas-KP/DSC00582.JPG - not sure why && not repeatable, so its not the image, timing/race condition maybe?
BUG-137: after moving/refiling photos, the next shift-click is out of order (reload fixes it)
BUG-138: Placeholder for all the ways we can get the front-end confused:
---> JUST fix all these BUGs (relating to confused/lost state) by revisiting the overally complex way I remember state and my position in a list (probably FAR easier, to make an initial sql just save all eids, and then not try to recreate that list ever again and not care how I got into the list). Can attach a "running server-side sequence number", and if old sequence, and the original eid list results in a failure, then just pop up that the saved list is no longer valid, and ask user to re-do their search/list..."