Fixes comparison with min_quantity orderpoint in scheduler - basic floating
point math issue in procurement scheduler when comparing current quantity
with orderpoint minimum quantity. In certain cases floating point comparison
could result in e.g 400.0 < 400.0 == True due to typical floating point
comparison issues, as Odoo doesn't use Decimal types where the issue
doesn't exist.
Fixes early exiting out of the loop cycle, in case qty is already near zero.
Fixes the new procurement creation check, to not do that if it's close
enough to zero already, to be considered a floating point math error, not
really non-zero.
These combined (or at least the last one) avoid each supply_method == buy
pending in draft PO's getting a zero quantity extra procurement order each
time the scheduler runs. Otherwise there could be hundreds of zero quantity
procurement orders pending, which makes the confirming of the PO take hours,
due to creating hundreds of stock moves for each order line.
Use float_compare helper to solve all these with floating point type for now,
instead of the more evasion possibility of converting to Decimal module.
Two potential bad comparisons remain, add FIXME notes for now until further
analysis.
Also: Float rounding on reste when comparing and on the procurement qty
* The dashboard from the backend howto doesn't specify "name" field
which causes non-null constraint violation.
* Some useful examples uses idea.idea" model which isn't available in
8.0 so I replaced them with "note.note".
closes#3993
The same is done when extra moves are generated. It is going to check if the UoM of the operation is smaller if it has one.
Throw an error when a key can not be found in action_done because there were links on a move
that was not supposed to be done (e.g. 0.5 Dozen when Dozen is rounded at 1)
[IMP] Throw an error when a key can not be found because of UoMs/picking + extra float_compare
[IMP] Integrate remarks qdp
[FIX] Remaining qty should each time be in the default UoM of the product
Even with different UoM we want a consistent matching between moves and pack operations.
When calculating the remaining qty on move/pack operation we always start by converting the
qty on the move/operation to the default UoM and afterwards we subtract the links between them
which will also be in the default UoM of the product.
In order to create backorders / extra moves these quantities are used.
When a negative quant is created but the positive quant counterpart is reconciling
a negative quant that of course also has a positive counterpart, the latter should eventually
let its field propagated_from_id tell that it originated from the very first negative quant as the
second negative quant will have disappeared through reconciliation.
Used to be the first line was the CSV headers, the slice was left over after
these were removed from the source data. It probably didn't hurt (only issue
would be if the first module — alphabetically — has a single translatable
term), but it's just as clean not to have that.
Also removed now-unused variable (probably leftover of the CSV thing as well)
This is to help forum moderators to fight against
spammers. It was previously difficult as the spammer
profile became unreachable as soon as their karma
went below 1, even if they had other questions
or answers still published.
It looks there is a bug in Firefox concerning responsive images in table. See bugzilla https://bugzilla.mozilla.org/show_bug.cgi?id=975632
Bootstrap advises to use width: 100% for .img-responsive as workaround were needed.
The @moz-document is to apply this for Mozilla only.
opw-617582
opw-618659
Mart Raudsepp