The last week and a bit more have been insane in terms of being away from writing - be it code, or a note. So, amidst a long “get to know the product” session from the founder, sat down improving the GraphQL sanitiser. Added a couple of changes from the earlier work that went kaput thanks to a ‘well thought out’ rm -rf on my desktop. First, an opt-in extension on the Express JS router to provide a simpler way to expose data schemas for any controller-model pair. Given it is opt-in and a question of data classes being exported by a model module, flexible enough for generic usage. ...

The last few days have been a slump - both in terms of productivity and writing. A few steps, must be acknowledged, are working thereof though. A five day hiatus now coming to an end with a bulk of the pending ones on the plate being sorted. But that does also open up the next phase of planning and such. While it cannot be in certainty told the systemic issues might see a solution soon, they are at the least out there in the open. That can provide a path towards planning, or damning, as one might find fit. ...

Checklists on paper are underrated and fancy tools, almost always, overrated. Back to jotting down on paper JIRAs that need action on priority, dependencies that need follow-up, and tasks for self that need focus time. One good thing about working with hand-written and crossed out lists is timeboxing that happens as a side-effect - a visible fixed list means, implicitly, work is siloed.

Allen Holub made an interesting quote during the course of a talk that struck a good chord. Scrum is basically a wrapper put around Extreme programming in order to make it more palatable for management. One of the initial discussions I had with our QA lead a few months back was to put in a discipline of all stakeholders discussing pre-sprint. A la a CEO pitching their making-the-world-a-better-place product and VCs taking their time to get back with feasiblilty and PMF questions, product owners pitch and technical stakeholders assess and get back. He said, well, that’s XP there. As a rose by any other name person, as long as it pushes things ahead. ...

One of the concerns about a water body like a dam is how well maintained the sluices are. And one should add to it, how well the pressure safety valves are engineered. Encountered yet another of our current stacks’ concerns today - a singular GraphQL-esque end point with no isolation. Add to the mix, a not so well thought handling of connections and we had a good CPU spike. Second day in a row dealing with database concerns isn’t a welcome Independence day. ...

Inching in some sense towards a freedom; another I’m trying to free is our APIs’ hackiness. Spent a bit on a vlog series that focuses on a new OS/language development. A couple of ideas to provide a slim, robust layer for GraphQL/REST APIs to restrict queries/mutations. Smaller steps from the past looking back.

The spatio-temporal bounds of perception and thereupon and thenceforth of inference make the two invalid modes of cognition and of assertion insofar the dominion of matters such as god are concerned. Whilst one might on the other hand employ the same limits to negate the presence, or of absence of divinity or such ideas mental or otherwise, the quandary in such is one of a non-sequitur if one may, for reasons stated above. ...

For most of my POCs/experimental work, I use my own [HTTP server package] (https://github.com/techiev2/node-http-server) and a scaffold bash script that boostraps a complete API including base tests, package.json, and git init. Comes in super handy so made a change/addition to this a few days back. Since the scaffold gets the base commits, setup done and I create the repo in parallel to continue on, added a default route that does a git pull on the repo. An unseen positive here is the fact that it is deploy-first an approach in a sense, since I need webhook end points. ...

A few days back, I wrote about something I built for personal use - a quasi ORM that depends on the database for typing and validations. While it is at a 0.9.x state right now (I’d like to update the API a bit more intuitive, ergo), found this on Hackernews a while back that takes a similar approach to things. The API update I’ve been working on for the last couple of days looks at using Tagged template literals to provide a flow similar to a GraphQL query. ...

A year on from this, a bit of an update on this; bandwidth is a great concern, without a doubt. Most of the changes came from a switch to MySQL from the MongoDB land. One of the questions I’ve had since January was working with type safety in case of NodeJS. TS/others do provide some solutions. The major concern though is runtime safety, particularly when working with the database. We moved away from Mongoengine in 2015/6 to Pymongo and experimented* with two type check versions; one each from from my colleague and myself. And experimental for the same reason - runtime safety. ...