Expected and unexpected benefits continue to drop out of our recent triplestore upgrade (weeknotes passim). Performance continues to “meet the expectations of the internet age”, as we say around these parts, but we’re also seeing an unintended bonus or two. Questions for written answer, tabled in one system, are supposed to be replaced with their answered versions - from a feed from a different system. But that hasn’t worked reliably for years. Also, oral contributions sourced from House of Lords Hansard may be randomly assigned to the Green Carpet Contingent, for reasons no one ever pretended to understand. This week, our Jianhan happily reported that both bugs are now squashed, again for reasons no one pretends to understand. Still, we’ll take every win we can get.
Success breeds success, and our Jianhan has acquired quite the taste for upgrading things. This week, he’s been casting an eye over our ORefFIS, which is supposed to read the rdfs:seealso properties in our triplestore and update identifiers for Members and Departments accordingly. A job it’s failed to do for perhaps ten years. A hard-coded bug, not difficult to fix, he reports.
Never one to work on one thing when he could be working on three, Jianhan has also been carrying out an archeological exploration of data.parliament. The poor old thing frequently slows to a halt, which given it’s a critical component in the publishing of Research briefings, is something of a problem. Our poor researchers can spend hours down the pixel mines, battling the briefing template, arriving at the point of pressing publish. And. Nothing.
Emerging with a face full of soot and a slightly bent computational spanner, he cast the room a quizzical look before asking, “who the hell built this?” We don’t know, Jianhan, we can’t say. Getting past his bemusement, he’s managed to export each and every triple from the antiquated store and load them all into the lastest and greatest version of GraphDB. Jianhan reports that there’s some script running that churns out a reverse chronological list of everything in the store, several hundred times a minute, and, as the volume of material grows, failure is only to be expected. In much better news, the upgrade appears to have solved the problem, all scripts running smoothly and taking considerably less time to do so. At least in the test environment. Stay tuned as we attempt to replicate Jianhan’s genius in production.
Upgrade endeavours do not stop there. Another of our ancient antiquities is also on its last pair of legs, this one being our thesaurus management application, confusingly named Ontology Manager. Like much of the rest of our software, it’s several major versions behind what’s considered contemporary. Once again, Jianhan has managed to export all the data and reload into the current version. Our crack team of librarians have tested and declared themselves happy. All well. All good. The problem is, the old version was based on Zthes, whilst the new version follows the SKOS specification, their respective APIs following the same convention. Which would be fine, if our API dependency graph didn’t look like overcooked linguine.
It feels like we have a couple of options here: either upgrade all the dependent software in one big bang, or find some way to emulate the old API from the new one. So far, Jianhan appears to favour the former, whilst Michael is a big fan of the latter. Then again, Michael doesn’t have to do the work, so his vote doesn’t really count. Luckily, Delivery Manager Lydia set up a call with the vendors who are currently looking into whether the old API can be emulated from the new software. Which would save everyone a great deal of time and indeed trouble. We wait in hope.
In the meantime, it remains safe to say that there is no shortage of work here. It seeming unlikely that Jianhan will ever find himself without a job. Should he be good enough to stick around.
Having made it this far, we fear we may have skipped over the question that is obviously on our dear reader’s lips: have you found a developer for your search application yet? Dear reader, we have not. Nor - despite Delivery Manager Lydia’s best efforts - does any such thing look imminently likely. Never ones to sit idle, your regular correspondents, Anya and Michael, have volunteered their time to help domain model the library publication process. So far, they’ve covered off bits of POST, and a couple of bits of the Commons Library, namely the good folks from the Parliament and Constitution Centre, and the Economic Policy and Statistics Section. So far, we haven’t quite got as far as anyone actually publishing anything, concentrating instead on the ‘commissioning process’. But it’s all been quite fun. Our thanks to Lorna, Chris, David, Matt and Oli(ver). At some point, somehow or other, we think this all stitches into our Single Subject View of the Librar(y/ies) work. Or at least, we hope so. More of which, almost inevitably, later.
Way back in 2024, a request flooded in from the Secondary Legislation Scrutiny Committee asking if it might be possible for our made negative statutory instrument procedure map to be adapted to include the laying - and potential withdrawal - of reports about NHS pension schemes laid under section 22(2)(b) of the Public Service Pensions Act 2013. Being obliging types, we replied in the affirmative. This week, Librarian Ayesha found the time to add five new business steps to both map and data, with instrument creation and report laying steps being actualised in all five pertinent work packages. Report withdrawal steps wait in reserve, should such a thing ever be seen to happen. Fine work, Ayesha.
Still with the SLSC, it came to our attention, that the labels applied to our reporting steps did not quite match the labels used in their reports. This has also now been remedied.
Happily, we’re also told that the SLSC have agreed to have their reporting criteria - from perceived inadequacies in the consultation process to insufficient information in explanatory material - captured as the usual set of steps and routes. The effects of which can be seen by both the addition of six new steps to our SLSC component procedure map, and also in our Procedure Browsable Space™. As ever, the latter will not only make their instantiation available as pixels, but also as RSS. Meaning an alert to any or all flavours of report piped straight to your inbox. Should you be in the - admittedly restricted - market for such things. For now, only a handful of recent instruments have seen these steps actualised, a separate card being created to backdate efforts to 2017. Only 431 to churn through, we’re told. Best of luck ‘brarians.
In procedure mapping adjacent news, Librarian Jayne continues to make fine progress on what the world now knows as our Procedure Browsable Space™. Aided and abetted, as ever, by Computational Mousemaster Michael. As he’d like to be called this week.
Fresh off the printing press are new pages listing enabling legislation and the work packageable things exercising powers delegated by an item of enabling legislation. The making of these pages has caused us to once more question our delegation ontology, which appears to suggest that all Acts of Parliament are a subclass of enabling thing. Neither the ontology nor our aging memories are clear as to whether we meant enabling thing to be a thing that delegated powers or a thing whose delegated powers had been exercised. Quite clearly, not all Acts of Parliament delegate powers, and, where they do, not all of those powers are exercised. We’re hoping we can find some way around this impasse without having to create a subset of Acts which delegate powers. That work being, quite frankly, beyond us. That said, we believe that’s a problem the magnificent Mr Sheridan is working on, so maybe he can come to the rescue at some point. In the meantime, a trip back to the whiteboard beckons. Until we reach the other side of that, our enabling legislation lists remain, as they say, not nothing. Who knew, for example, that the European Union (Withdrawal) Act 2018 has, so far, enabled 1,236 laid statutory instruments and statutory instrument-like things? One for the pub quizzers amongst us.
That work means we also now have lists of enabling legislation for things being enabled. The Social Security (Contributions) (Amendment No. 5) Regulations 2025 for example, being enabled by both the Social Security Contributions and Benefits Act 1992 and the Social Security Contributions and Benefits (Northern Ireland) Act 1992. It remains slightly sad that Linked Data - Upper Case - got a bit of a reputation as being the domain of beared weirdoes. It was, and remains, the golden path to toward functioning hypertext. As we’re sure our dear reader will appreciate.
Our only other bit of Procedure Browsable Space™ news is the addition of a list of organisations accountable to Parliament, the work packages their layings have resulted in, and the treaties for which they have been the departmental lead. All lovely stuff, though the dates assigned to said organisations look fishy to our eyes. The Department for International Development, for example, having a creation date of the 1st January 1900, rather gives that game away. For that reason, those pages have moved, not to the happy pile, but rather back to the “blocked until we find out what’s going on pile”. Librarian Jayne continues to investigate.
It should be stated, if it were not obvious, that our Procedure Browsable Space™ is very much a Phase One Endeavour. For now, we’re concentrating efforts on giving all the things URLs - URIs if we must - and making lots of lists. Once that part is done, we need to think about who the potential users are and how to make it useable. We’re currently supporting views of the data that even Librarian Jayne would struggle to find useful, a list of all the routes in all the procedures being the most obvious example. All the views will stay, but the ones we provide prominent links to will change - user needs being emergent from materials, as it were. It might not be how everyone makes websites, but it’s how we make them.
Luckily, Mr Korris has agreed to come aboard as our House of Lords advisor. Thanks, as ever, Matt. Should our dear reader find themselves curious and of the Green Carpet Clerkly Contingent, please don’t hesitate to get in touch.
Fans of our beloved egg timer™ will be delighted to learn it’s currently undergoing something of a refresh. Of primary concern is the information architecture, which has grown somewhat organically, resulting in bits that stick out like the proverbial dog’s leg. Whilst we’re on the job, we’ve also taken the opportunity to paint it in more parliamentary appropriate pixels and rebrand it to something that sounds vaguely more professional than the egg timer. Though what that might be remains under discussion. It will always be the beloved egg timer in our hearts. There is nothing to show as yet - well, there is, but it’s all on a test server and none of us want the web seeing that - but, maybe in a couple of weeks or so, the great unveiling will take place. And the world will be appropriately stunned. Perhaps.
Of more immediate import, Shedcode James has rolled out the calendar synchronisation fix to live. We’ve had a long-standing problem whereby the syncing script worked perfectly, until the moment something in the calendars it was synching from changed. At which point it either worked or fell over. It mostly worked to be fair, but mostly is not really good enough. James had already added alerting - or observability, as Young Robert would no doubt say - so we’d moved past the point where Librarian Jayne was forced to check the syncing logs every day. Now he’s gone a step further, adding recovery code should the syncer once again fall over. Librarian Jayne has attempted to test, but the blasted thing won’t break. And we’ll only truly know if the problem is fixed if it breaks. Typical.
Following a last minute hoo-hah over the valid vote count - apologies Statistician Carl - we’re delighted to report that the results of the Runcorn and Helsby by-election have now been verified and uploaded to our many and varied systems. Way more systems than one might reasonably expect and more fine work from Librarians Anna, Emily and Phil.
Librarian Emily has also been tidying the spreadsheets that accompany the Library’s general and by-election briefing pages, consolidating constituency names, party names and county names, whilst also applying a dollop of Democracy Club candidate identifiers for good measure. Marvellous.
In our final bit of psephological news, tables listing maiden speeches in a Parliament period are now sortable. Which we’re told will make constituency expert Neil’s life in particular a little easier. So not nothing.
The past few months have seen us hard at work, attempting to create a Single Subject View of the Library™. In the first instance, this takes the form of a replacement for the Library’s Subject Specialist Finder [more weeknotes passim]. In the longer term, we hope to see it encompass all of the Library’s inputs, and indeed outputs. At this point, tradition dictates that we go into mind-numbing detail of what has been achieved and what remains to be done, but the Subject Specialist Finder contains a small amount of personal data, so is placed firmly behind locked doors. Mostly, you’ll just have to trust us. Let’s just say that this week Librarian Susannah circulated work so far to the Library’s aforementioned subject specialists and we wait, breath bated, upon their response. It is never easy to demo anything to people who are personally invested in the subject, made doubly difficult when the demoees are the subject matter at hand. Please pray for us.
In some attempt to get ahead of the brickbats, Librarians Anya and Susannah saddled up their computational donkeys - Young Robert and Michael - for a trip to the seaside and a meeting with Silver. Which is how Friday found our famous fivesome attacking a whiteboard with dry wipe markers in an upstairs room in Brighton’s Friends’ Meeting House. And where better for such good friends to meet. Whiteboard suitably graffitied, the results have now been decanted into Trello, roughly categorised into things we know we need to do to make the data ingest better and things we know we need to do to make the data output better. Which we think covers everything we know to be wrong and, hopefully, a fair smattering of what the subject specialists will find to be wrong. Everything under our control once more feels under our control, the slight matter of how we take this toward production remaining something of a mystery. Please pray for us.