Friday, September 8, 2017¶
No alternative for appy.pod¶
I did some more research for:ticket:1534 because for Lino Avanti we want to print documents.
I installed unoconv on my
sudo apt install unoconv) and tried it. I had to specify
UNO_PATH, otherwise it worked out of the box:
UNO_PATH=/usr/lib/libreoffice unoconv -f pdf myfile.odt
But unoconv does only the smaller part of what appy.pod does. The actual benefit of appy.pod is the templating work, i.e. how to replace “markup codes” in the template by data from the database. Secretary is one alternative for this, and there is also py3o.template. But after looking at these once more, I continue to be skeptical. appy.pod is by far the best.
Filtering contracts in Lino Welfare¶
I worked on #2030.
Dans les listes de contrats (PIIS, Art60, Art.61 etc), Lino permet de filtrer sur une période donnée. Si l’on remplit au moins un des champs “Date du” et “au”, alors il faut également spécifier dans le champ “Évènement observé” comment Lino doit interpréter cette période:
Actif : applies_from (Début de contrat) et date_ended (Date de fin) doivent se trouver dans la période donnée.
Commencé : applies_from (Début de contrat) dans la période
Fin : date_ended (Date de fin) dans la période
Décidé : date_decided (Décidé le) dans la période
Issued : date_issued (Date de signature) dans la période
Note: il n’est donc pas possible de filtrer sur le champ applies_until (Fin prévue), sauf en utilisant un filtre de colonne.
I changed the order of items in
ContractEvents: active is now first because it is the default value and the most used.
I changed the meaning of ContractEvents.active : it now ignores applies_until. This behaviour was from a time where end_date was possible empty (which is no longer true
This is defined in
I replaced applies_until by date_ended in the default column_names of the Contracts tables in jobs, isip and art6
Filter list of clients when creating appointment¶
Getting Travis projects to build¶
django-admin.py prep: error: unrecognized arguments: –noreload
That was because yesterday I removed the
--noreloadoption from :manage: initdb. I removed it because AFAICS it was useless. So now I also removed it in the inv prep command defined in
No .coveragerc file in welfare
The Lino Book currently has 59% of coverage.
Together with Johanna we found a way to break user management: create a user with all of the following fields blank: first_name, last_name, username and initials. Fixed.
The site of a ticket¶
lino_xl.lib.clocking.MySessions we had a remote field
ticket__project. This field was useless now, I replaced it by
ticket__site. Because before sending out service reports for August
I wanted to verify that I didn’t accidentally forgot to assign the
site of a ticket I have been working on.
This change caused Lino Presto to break. Since Presto has never been used and has no value, I deprecate the project and removed it from my list of atelier projects.
The model of a VirtualField¶
While experimenting in Exploring SQL activity in Lino Noi I realized that the
model of a
VirtualField must be the class where the field was
defined. This can be an abstract model. The VirtualField instance
does not have a list of the concrete models which use it (because they
inherit from that class).
was doing the following:
for model in models_list: for k, v in class_dict_items(model): if isinstance(v, fields.VirtualField): v.attach_to_model(model, k)
And meth:class_dict_items was defined as follows:
def class_dict_items(cl, exclude=None): if exclude is None: exclude = set() for k, v in cl.__dict__.items(): if not k in exclude: yield k, v exclude.add(k) for b in cl.__bases__: for i in class_dict_items(b, exclude): yield i
That was incorrect because for virtual fields defined on a model mixin we must differentiate the “defining” from the “using” model.
Release in Chatelet¶
Since 2017-05-10. Upgraded Django from 1.9.13 to 1.11.5.
TypeError: ‘profile’ is an invalid keyword argument for this function
(because field User.profile renamed to
Memory usage on saffre-rumma.net¶
Maybe Jane is slow because saffre-rumma.net had only 276M (of 2G) free memory?
$ free -h total used free shared buffers cached Mem: 1.9G 1.6G 276M 39M 30M 193M -/+ buffers/cache: 1.4G 500M Swap: 1.0G 376M 647M
After a reboot it was 1.1G. But after the first web requests to jane and bugs these 1.1G were reduced to 574M.