For once, it would appear we have made good on our promise. Two sets of weeknotes in two weeks being quite the treat for our regular reader.
Progress on new, old search continues to be made. At pace, as Young Robert might say. At scale, as he might add. Glancing across at the burndown chart on our interactive dashboard, we notice only a slight blip. This due to Developer Jon suffering under the weight of a cold of catastrophic proportions. The last time we saw the lad, he reported that he was “dying of death” and promptly went back to bed. Get well soon Jon. We need you.
Work is currently focussed on tidying what has gone before, though Librarian Anya has taken to ‘grooming the backlog’ into what Young Robert might call ‘epics’. Quite a number of ‘epics’ still waiting on the todo pile. In tidying news, the leading of text in result items has now been fixed, research briefings lacking ‘landing page text’ now fall back to description text and an answering body label has been applied across questions, both written and oral. Reflecting changes made elsewhere, deposited papers now have a single author field encompassing both corporate authors and personal authors.
Unfortunately, Jon is not only suffering under the weight of his man-flu affliction, he’s also reached the end of his contract. We found some money, so he’ll be sticking around but not for the full working week.
Whilst we have no new maps to show you this week, Librarian Jayne has updated pretty much all of the old ones. This to say that, should an instrument be laid and not withdrawn, certain committees are caused to report on it. Whilst an instrument being laid and withdrawn changes the causal relationship to a more appropriate allows. No one is taking away the right of a committee to consider a withdrawn instrument, the decision is theirs and theirs alone.
This late breaking change arose because - when we remapped all our procedures into their new and improved logical and arithmetic shape - our design patterns said a Decision Step must sit immediately before a Business Step, AND and OR steps being binary and unable to process an input of ALLOWS. A late change to ternary logic gates - weeknotes passim, let’s not go there - means ANDs and ORs have expanded their input and output states by a good 50%. Our procedure parsing code has been updated accordingly, conditional switching from causes to allows becoming possible in the process. Which is probably way too much information, but we never like to hide implementation details from our esteemed reader.
Still on the subject of procedure mapping, what Young Robert might call our procedure tracking ‘stack’ relies on three components, all somewhat neglected in terms of development effort. Sitting in the middle is data platform #3. Sometimes wistfully referred to as the Samu triplestore. For reasons we don’t pretend to understand, this continues to work like clockwork, despite having been barely touched by human hand for the past seven years or so. That said, any data platform is only ever as good as the tools provided to get data in, and the tools available to query and serialise the data back out. Let’s start with getting data in.
One of the things sitting behind data platform #3 is the procedure editor application. This is where our crack team of librarians manage both the procedure map data that we all know and love, and the tracking of individual events during Parliament’s scrutiny of instruments subject to those procedures. It was intended to be an interim solution, the fact that it sits atop a database called the ‘interim database’ giving that game away. Unfortunately, it’s now been an interim solution for the last seven years and its feet are on their last pair of legs.
There are two main problems with procedure editor:
The data model is locked in a relational database, any changes requiring developer effort to both schema and application code. Where developer inevitably means our Jianhan. Our Jianhan being nothing if not time poor.
It is heavily reliant on JavaScript with no pagination. The code to get a list of things brings back the absolute bare bones, which then get decorated by what people used to call AJAX, but now just call bad design. This means that, as the volume of data increases and the lists get longer, the value of n in n+1 tends toward infinity. HTTP not being as young as it used to be, it gets a little tired and some pages now fail to load. And there’s the constant threat of the thing just turning up its toes one day. Being resilient types in the face of computational adversities, our crack team of librarians have developed workarounds, but it’s all less than ideal.
For this reason, Librarians Ayesha and Jayne have dragged computational whizzkids Young Robert and Michael into a series of meetings with family friend Silver. Their aim, to investigate whether it might be possible to port both model and data into Data Graphs. There is no promise of success, but progress is being made. A ‘safe to fail probe’ then, as Young Robert might say.
This being an admin application, there’s nothing we can really point our dear reader at. You’ll just have to take our word on the progress part. That said, we’ve also taken the opportunity to rework some rather core parts of our procedure ontology, which is now looking better than ever. In our opinion. Though we would not advise looking too closely. Whilst we’re more than happy with the shape of the thing, the comments could use a little work. That’s waiting on the return of Robert, and his tiny commenting chisel, from his ‘golfing vacation’.
If you were unfortunate enough to be assigned weeknotes reading duties last week, you’ll know that we’ve also made some progress on the other end of the pipes. Whilst our statutory instruments and treaty tracking websites are more than fine if you want to know what things have happened to an instrument, they are less fine if you want to know which instruments a thing has happened to. Librarian Jayne’s SPARQL library bearing testament to unmet user needs - from both inside and outside Parliament - in this regard. For that reason - and to give the SPARQL library a more comfortable home - we’ve also started work on a thing with the working title of procedure browser. Which is also going remarkably well. As ever, we’re starting work by establishing the bone structure and designing from the domain model outwards. After all, why get distracted by pixels when there’s the joy of URL structures to get stuck into? As of this week, we have all pages under step types, steps and step collections coded and ‘complete’, with the merest hint of the actual procedures making their presence felt. If you drift down to the bottom of any page, you’ll also find the SPARQL queries used to construct that page. Thanks be to Librarian Jayne. Being firm believers that lifting the bonnet should always be possible, it’s the bit we’re most chuffed with. Obviously.
Replacing procedure editor is not our only Data Graphs project. We’ve also been looking at it to replace our Odds and Sods Information System (OaSIS) - a system to create records for things lacking feeds - and to provide the underpinnings of our Single Subject View of the Library. The latter being a project to place everything the Library knows about a subject - specialists, briefings, insights, current awareness, enquiries and the book catalogue - in one place. This to make life easier for both researchers and customers.
Way back in week 37, Librarians Anya and Susannah took their computational lapdogs, Young Robert and Michael, for a stroll along Brighton prom. Miniature train travelled on, they met with Silver for a Single Subject View planning session. Now Michael - as you may know - has never been a big fan of what people tend to call ‘design deliverables’, believing that if a service is intended to be delivered via a browser, it should be designed in a browser. Nevertheless and under sufferance - but without whining - he took out his sharpie and a fag packet or two and sketched out some wireframes. This week, we saw the first fruits of said ‘design deliverables’, Developer Charlie turning Michael’s fag packet sketches into actual working code. And lovely it looks too.
A Single Subject View of the Library is, of course, only made possible by the love and care that has been poured into our taxonomy over the last 40 odd years. This week, that taxonomy flexed once again, when the committees website team recategorised all joint committees with ‘draft’ in their name as Select Committees. Said committees were also subjected to a - website only - rename, the ‘Joint Committee on the Draft [whatever] Bill’ being restyled as the ‘Draft [whatever] Bill (Joint Committee)’. We can only assume that someone, somewhere had a listing problem. The taxonomy has been updated to reflect.
Last time out, we reported that Shedcode James thought he’d fixed the lack of paragraph numbering on our standing orders application. This week, we’re delighted to report that Librarian Claire has checked that work and confirmed it is indeed correct. It being ‘a delight to use’ coming as an additional bonus. Top work James. Top work Claire.
This week has seen outreach and engagement efforts both cross-border and closer to home. Librarians Emily and Susannah accompanied boss boss ‘brarian Bryn to Edinburgh, to attend the Inter-Parliamentary Research and Information Network annual conference. We’re told that Susannah and Bryn took to the stage for a workshop on preparing for and dealing with a general election, co-presenting alongside a researcher from the Senedd, We’re also told that our election results website got a shout out. Which is always nice. Emily reports that the Library’s very own Sarah Priestly gave an excellent talk on supporting cross cutting research and that colleagues from POST were also in attendance. We would link you to the conference website but it is, apparently, under construction.
Michael also managed to do his bit, delivering a talk on the philosophical aspects of triplestores to computational section colleagues. Librarians Anya, Ayesha, Jayne and Phil popping across to Victoria Street for much-needed moral support. Open world models were mentioned, closed world models got a look in, inference and reasoning were both explored. It seemed to go quite well, Michael managing to not trip over his own tongue for the best part of 40 minutes. A personal record perhaps.
