Friday, February 19, 2016

I discovered Atom. There are chances that this is going to replace Emacs for me. It took me just a minute to get Alt-Q configured to do what I want. (This was one of the things which made me not switch to PyCharm when I last tried it).

Alexa discovered that eiche2lino did not correctly import all fields. The reason was rather difficult to believe: xlrd has appearently problems when reading cells which contain formulas of style [=TRUE()] and [=FALSE()]. These display correctly as 0 or 1 (or FALSE or TRUE depending on the application which views them), and xlrd converts them to 0.0 or 1.0 float values, but [=TRUE()] does not always give 1.0, and [=FALSE()] not always 0.0. I worked around this by modifying my input file: Select the concerned fields, Ctrl-C, and then use “Paste Special” to copy only values, not formulas. About 90 minutes of work!

The number of participants

Alexa would like to see the total number of participants of a course. Actually the total of the lino_xl.lib.courses.models.Enrolment.places column. Unfortunately Lino does not show totals in an Ext.grid. Fortunately it does show them in HTML and pdf view. But unfortunately both these views don’t work for EnrolmentsByCourse, for different reasons. I opened ticket #788 for this composite problem.

Printing the table to pdf fails due to our good old friend #620. I am undecided whether this ticket is worth working on it. It looks tempting to reimplement the pdf view using wkhtmltopdf in order to avoid the problems of html2odf. OTOH appypod remains important, e.g. for users who want Lino to generate editable documents (.rtf).

Printing the table to html works, but it does not respect manually added columns (#789, HTML view of a table ignores manual modifications to column layout). And EnrolmentsByCourse currently does not show the places column. So even if you unhide it manually, the HTML button still won’t show you the sum of your places.

Side effect: I moved the .odt templates used by lino.modlib.appypod.mixins.PrintTableAction from lino.modlib.lino_startup to lino.modlib.appypod.

Meanwhile, as a workaround for Alexa, I started to make a new printable “Overview” document of a course. A new template overview.wk.html in the config of lino_voga.projects.roger.lib.courses is a first draft for this document.

In order to get the sums in this document, I added SumCollector to the default context of a Jinja template in lino.modlib.jinja.renderer, and I optimzed a bit on the API of that helper class. I am not yet really proud of that API, though. And oops, later I noticed that these optimzations caused a failure in Lino Welfare so that I had to undo them (partly).

Lino XL, the Lino Extension Library

About #356 (Split Lino into “Core” and “Enterprise”).

In a short brainstorming session with Hamza we found the name for the that new Lino repository.

Hamza pointed out that lino_erp or “Enterprise” is wrong, because Lino Così , Lino Voga are also “Enterprise” functionality.

For example lino.modlib.contacts and lino_xl.lib.cal should certainly go into this new “extended” Lino. Because if somebody wants to write her own implementation for contact management, then they should be free to do it. They would use just Lino Core and not Lino extensions. Only the “System plugins”, “User interfaces” and “Utilities” (http://lino-framework.org/api/lino.modlib.html) should remain in the core Lino repo.

So the name will be Lino XL (“Extension Library”), Python name lino_xl.

Another question was: A clone of the Lino repo is currently 120 MB, most of which is “useless” history. Should we use the occasion and create two completely new repositories? This would remove history from the working codebase without removing it completely from GitHub. Pro: an efficient and quick solution. Contra: Makes project history less transparent and statistics more difficult. Actually the size problem should be solved by providing source releases.

And actually, when the XL will be moved out of the core, this code will indeed become rather stable and it makes sense to use officially released versions on production sites.

Tolerating Sphinx warnings

I also continued with the long-term project of eliminating all Sphinx warnings from my blog. For this I tempoerarily change the tolerate_sphinx_warnings from True to False. But it is too early to see the end…