The water heater aka boiler croaked... umm, that can't be, it's still under warranty, these things are made to croak at least a couple of month after it expires... Let's see... if the heater blew, we wouldn't have noticed because it has two [of them]. Well, the glimmer into hand and see where it has power. Mmmm... has it nowhere, so it's the power supply. Let's check its fuse... aha, blown. Turned it on again, blew right away. Okaaay...
Turns out the cable croaked... that is, not the cable as such, it's the point of contact where the cable was extended, in this box. It seems there was another burnout like this before, so the charred end was trimmed, which made the cable too short, so it was extended by two pieces of wire of less than 10cm in length. Okay, the contacts were done with these caps which screw onto the wire twist, it holds but it's still open. When I took those off, well, a sight to see - the cable to the fuseboard was aluminium... Awwww fuck, and this is the little bathroom. Excellent combination, in a humid room (though this is actually quite dry, as bathrooms go), copper and aluminium touching. Good [thing that] we didn't burn.
Today we replaced that cable from the fuseboard to the boiler. Luckily, it was nailed to a beam in only one place, which I could reach by hand, through yet another hole we cut in the ceiling. Goodbye aluminium, long live copper. From that tiny bathroom to the fuseboard there's less than three meters, and the ceiling is hollow anyway. Went to Loos (okay, Lowe's, but who gives it a fuck, we call it so, the pronunciation here being a matter of inspiration and stabs in a dark), and took a cable one gauge thicker than needed, Justin Case approves that.
We found out that the boiler also had some launcher, a piece of electronics through which it's powered, which would be that larger box on the wall, but it was unplugged long ago and the power goes straight in. Guess it was disconnected then the copper-aluminium connection burned the first time.
The car was at the repairs, the guy drove it here when done. Tomorrow we should visit Go, and the bearing in the front right wheel is becoming louder, so I reckon we better fix that before the trip. It works like that for two years already, but, again, Justin Case is my friend.
On twonyninth I picked on something for Sean, a new registration key generator, now includes the customer's school in the keys table. While on it, I noticed that we're using MyEscuelle for something, now look at that, forgot completely. Did even that... To confirm, on UA there's this:
According to a survey Microsoft did, 50% of VFP developers use SQL Server.
And that depends. In all the apps I've done in the last four years, four have used SQL server, one is in the migration process, and one uses MySQL for some tracking - the rest (in that app and elsewhere) is just plain old DBFs. And I don't know how to count them, because some overlap partly, some had a dbf version running in parallel, one used SQL only on the side while the main work was in dbfs etc etc. The border is murky.
I figure "50% of VFP developers have used SQL server at least for one app" would be a more correct interpretation of the poll result.
Well, fuckit, this is the app for law students, so half the wisdom is in the registration, to make sure that they pay and until then that they may play but not do everything. Specially the handling of stevka (v. house dictionary), to limit it to work inside the app but erase it on exit. And it's only getting more complicated, he always needs one more thing, another option, another check... when I see what this one registration form is doing, well fuckya it's worse than ŠV-20. And to this he said „remove that validation, I didn't ask for it“. Well, I said, „I didn't put it there, it was there in the code since dark ages“.
On the other front, David sent some instruction to his folks that they have to have the pkzip25.exe in the directory where the Nyork backup is to be unzipped or else there'll be trouble and Jüzek's brother will be pissed off. By next day it worked, the version reported in the log was 1.0.12, and it did a bunch of stuff, ground and grinded some excel reports, then disseminated (!) them, tried to email some, of which most things worked and a couple didn't, one recipient didn't get the email, one report wasn't generated, but it trudged on and reached the end. The error processing I did old school way, as I learned on the eHosp, the code must not stop, may only skip the parts it can't do. And really, one of his guys said „it still shows some errors but I don't know how important they may be“. Well they aren't, one report one email, big deal.
On third front, Gary complains that the text he exported from the table, then messed with it in an editor, didn't import back. Tried the file myself, and it actually did okay, except it didn't find the end marker, something was missing there, and it didn't exit elegantly. And then he also managed to fuck up some delimiter line inside the text so the text for two records coalesced into one... but we solved that as well. We're also thinking of a spellchecker, i.e. to have a component, a simpler sideapp, which would enable his contributors to send him text in some technically clean shape.
Three days later, Go sent us a link to Gugao's maps, of a spot in Orlando where there's some chance of a job. Later she went there and had the interview, and I don't know how it happened that she had to walk last few blocks, guess she was using what public transportation they had, and this is when we acquired the remote navigation skills. She'd call us on her cell, tell us where she was at the moment and where she needs to go, we'd find these on an onscreen map, decide the best route, send her instructions. Worked every time, now and at a few later occasions.
21-X-2024 - 14-V-2026