SIAG-LA site design 2014

This document describes the re-design of the SIAG-LA web site as of early November 2014. This reflects some changes that I'd intended to make since the start of my term as secretary of SIAG-LA; I hope these changes will make it simpler not only for me and my successors to maintain, but also for others to propose contributions.

General design

The current site style is built on Pure, a small set of responsive CSS modules from Yahoo. In this setting, "responsive" means that the site should adapt gracefully to different screen sizes and window sizes. The new design hues closely to Pure's side menu layout example, and I have in no way modified the default CSS behavior from the Pure framework.

For content management, I am using Jekyll 2.4.0 a plain-text static site generator. Jekyll may be the most popular of the available static site generators, in part because it is used in Github Pages. The chief goal of using Jekyll is to separate the content from presentation, making it simpler to apply general stylistic changes in the future.

Jekyll has the option of using different Unicode encodings. My preferred default would be UTF-8, but there is some issue that I don't entirely understand that only appears on the SIAM server (vs local deployment or GitHub pages) that causes UTF-8 symbols to be mis-interpreted; so we're sticking with HTML entities for now.

The plan is to maintain the source code in a git repository on GitHub, which will automatically be re-compiled and uploaded to the SIAM site once a day.

Posts

Posts are stored in the _posts subdirectory, and are named according to the usual Jekyll convention (YYYY-MM-DD-topic.extension). The YAML header should include title, date, and author:

---
title: Title
date: 2014-11-06 19:32:45 -0500
author: David Bindel
---

Post excerpts are currently displayed on the main page in reverse chronological order, immediately after the introductory text. This will probably need to be paginated at some point, and perhaps it will make sense to automatically generate an RSS feed; but not now.

Conference records

By volume, the majority of the data on the SIAG-LA site relates to past and future conferences. It is good design in general to separate data from presentation, and particularly so with something like conference data that we want to keep up-to-date and to present in several ways. So I've created a _meetings subdirectory (a Jekyll collection) with one file per record. I format any text using Markdown, and follow the same convention used for posts (YYYY-MM-DD-shortname.md), where the date is the start date of the conference. This convention allows ordinary lexicographic ordering to coincide with chronological ordering. Each record includes the conference name, an optional URL, the start and end dates, the conference location, and the "type" of relation the conference has to SIAG-LA:

sponsored: Sponsored by SIAG-LA
cooperation: In cooperation with SIAG-LA
organized: Organized by SIAG-LA
siam-la: SIAM Linear Algebra Conferences
siam-la-biz: SIAG-LA business meetings
issnla: SIAG-LA summer school on numerical linear algebra

The type of the conference can be left blank to indicate that there is no particular relation to SIAG-LA other than that the conference may be of interest to members. The conference record YAML looks like

---
title: Twelfth SIAM Conference on Applied Linear Algebra
page: http://www.siam.org/meetings/la15/
start_date: 2015-10-26 12:00:00
end_date:   2015-10-30 12:00:00
where: Atlanta, Georgia
type: siam-la
---

The call for papers will be posted in December 2014.

Note that the dates must include a time field as well as a date in order for the template to work properly. This has to do with the way that Ruby handles the distinctions between dates, times, and strings; I tried to get the logic to work even if I just specified dates, but after a few minutes I just moved on. This could be improved.

The script meeting.sh in the _util directory automatically creates a new meeting record with the appropriate format.

The conference page has tabs that display summaries of all conferences and more complete information for a subset of conferences (upcoming conferences, SIAM LA meetings, summer schools). There is currently no way to see the full text records for arbitrary past meetings through the web page. This is not different from how the site worked before, but could perhaps be improved.

The templating system ensures that the formatting on these pages is consistent; it does nothing to help with correctness of the records. Ideally, if I've made data entry errors, someone will be able to call me on it.

Job postings

Job advertisements posted to the SIAG-LA mailing list are copied into the _jobads subdirectory. Job ads are posted in reverse chronological order by posting date on the jobs web page. Only ads that have not passed their closing date are listed.

Unlike conferences, job ads do not get their own pages. The reasoning is that a past conference is still likely of interest (and there are many of them), while past job ads are less interesting. The data remains in the repository, so someone could change their mind about this in the future.

The job record YAML looks like:

---
title: 
page:
posted:
closes:
---

If no closing date is explicitly mentioned, I default to two months after the posting.

Forms for job and conference postings

Job and conference advertisements can be submitted through email or via a Google Forms setup. The forms have an attached script to email submitters when they have successfully submitted an advertisement. The sync-jobs.py and sync-conf.py scripts under the _util subdirectory create appropriately-formated files for inclusion under the _meetings and _jobads directories. For example, to post a new conference ad submitted via the web form, one might (for example) do:

cd _util
python sync-conf.py
cp 2015-10-01-foo.md ../_meetings
git add ../_meetings/2015-10-01-foo.md
git commit -m "Added foo meeting"
git push origin

Note that the sync-conf.py script and sync-jobs.py script require pandas, gspread, and oauth2client. All of these can be installed via pip.

SIMAX papers

Information on recent SIMAX papers is automatically scraped from the SIMAX RSS feed by the _util/parse-simax-feed.py script. The current database is saved in YAML format in _data/simax.yml. The scraper runs automatically once a day; if there are any new papers, this will trigger a site re-generation. The simax.html template is automatically populated from the ten most recent papers, displayed in reverse chronological order immediately after the introductory text.

Unfortunately, Jekyll doesn't seem happy to convert UTF-8 to HTML entities on the fly, so the scraper script does the conversion (and saves the version with HTML entities in the database).

Prizes and officers

Information about the officers is saved in _data/officers.yml; information about the SIAG/LA prize winners and their papers is in _data/prizes.yml. Formatting information is given at the start of each file.

The prizes before 2003 were not previously listed on the SIAG-LA web site. This has been addressed to the best of my knowledge, but it took some digging to find the papers associated with the earlier prizes. Again, I hope that if I've made errors, someone will call me on it.

dbindel/siagla