Cast Iron Carl

Growing means active. Dry to the touch means intermittent, which fits a shower rather than a standing leak, and that narrows it usefully.

Before you cut anything, in this order:

• Pull the bath panel or open the access hatch and look at the trap and waste connection with a torch while somebody runs the shower for ten minutes. Most 'under the bath' leaks are the waste, not pipework in the wall.
• Check the overflow. A failed overflow gasket only leaks when water reaches that height, which is exactly the kind of intermittent that dries between uses. If your shower head is high and water runs down that wall, this is a strong candidate.
• Press the grout and sealant in the corner where the wall meets the bath edge. Soft or missing there sends water behind the tiles and out onto the ceiling.
• Dry the ceiling, draw round the ring in pencil with today's date, run the shower for fifteen minutes and look the next morning. If it grows past the pencil you have confirmed the source.

If all of that is clean, then yes, open a small square directly under the ring. A 30cm hole is a cheap repair. A ceiling that comes down on its own is not.

On the money side, be careful with used ones. The mattress part is essentially foam rails plus air chambers and it holds up, but pumps and remotes are the parts that fail and they are proprietary, so a replacement pump is a manufacturer part at manufacturer prices rather than a generic. When I priced mine a couple of years ago it was a meaningful fraction of what the whole used bed costs. Factor a pump replacement into your third of retail and see whether it still looks like a deal.

Also stop leaving fuel in it over winter, or use a stabiliser. Modern pump fuel with ethanol starts going off in about a month, absorbs water, and leaves exactly the varnish you are about to scrape out. Either run the tank and the bowl dry at the end of the season or fill it with treated fuel.

Do the fuel cap test before you take anything apart, just in case you get lucky. It costs ten seconds.

Cracks at the corner of an opening are the most normal cracks in a house. That corner is a stress concentration - the wall above the door carries load and the material either side does not, so seasonal movement shows up there first. Twenty centimetres, 45 degrees, one location, nothing else going on is textbook.

It keeps coming back because rigid filler cannot handle a joint that moves. What works:

• Rake it out into a small V rather than skimming the surface.
• Fill with a flexible filler or decorator's caulk rather than a hard powder filler.
• Better still, bed a strip of self adhesive mesh tape over it, two thin coats of joint compound, feathered 20cm either side, then paint.

The mesh gives the movement somewhere to go. Mine has held five years on a crack that used to reopen every winter.

What would change my answer: the crack widening past a few millimetres, the door starting to bind, or new cracks appearing on the same wall.

Filler is for holes, tape is for movement. Almost every repeat crack is somebody using filler where the wall wanted tape.

I would drop the pan sauce and put knife handling and heat management in the first two slots. A pan sauce is a lovely trick and he will make one about four times a year. Being able to cut an onion without hating it, and knowing that his hob has one setting that is right for browning and another that is right for everything else, is what determines whether he cooks at all on a Wednesday. Sauces are for people who already cook.

I own both and use both, and the honest division is that the slow cooker wins anything you want to fall apart in liquid, and the water bath wins anything you want to stop at a specific texture. Chicken breast is the one that converted me - it is the difference between edible and actually good, and no slow cooker will do it. Pulled pork in a bag is a fun trick and it is not better than yours.

Agreed on stews, and I would still take the pork shoulder in a bag over the slow cooker version for texture. But you are right that it costs an extra pan and forty minutes to get there.

Always. That photo has saved me twice on machines where the manual diagram was wrong.

Do not adjust the governor spring to mask it. Every surging generator I have bought secondhand has had its linkage bent by somebody chasing this symptom, and then you have two problems.

No argument with any of that, but note it applies to a VPS you administer by hand. A container on a managed host is $10-15 rather than $7, has no ssh, no disk to fill and no TLS to renew, and it is still a single region single process app that any of us could reason about. The gap between "box I patch" and "edge platform" has a lot of middle in it.

Do the arithmetic before the architecture. 3.9M/day is about 117M requests a month.

On the paid Workers plan that is roughly $5 base with 10M requests included, then 107M extra at $0.30 per million, so about $32, plus CPU time on top - at 5ms of CPU per request you are looking at another ten dollars or so. Call it $45-50/month. Check the current numbers yourself, but the shape does not move.

The VPS is $7 and 45 req/s of cached JSON is nothing for 2 vCPU. So the money says VPS by a factor of six.

The money is not the interesting part though. Your 70% cacheable is - stick a CDN in front of either option and 70% of your traffic never reaches your origin at all, which takes 45 req/s down to 14. At 14 req/s a $7 box is bored and the p95 question becomes entirely about the distance to Postgres, which is the same for both options because the database has not moved.

One person, no on-call: I would take the VPS, put a CDN in front, and spend the difference on a managed database and backups I do not run myself.

It usually is the thing. Most "do I need edge compute" arguments are actually cache-hit-rate arguments wearing a costume.

Nobody ever does, and then everybody has a door that swells in summer.

Do not plane it, and you have already worked out why.

Check the top hinge screws first, and not just whether they turn - take one out and look at how long it is. On a door catching at the top latch side, the classic cause is the top hinge pulling away slightly because the screws only reach the jamb lining rather than the framing behind it. Replace one screw per hinge leaf with a 65 to 75mm one that bites into the stud and the door lifts back square. Ten minutes, and it fixes far more sticking doors than planing does.

If that does not do it and it truly is seasonal swelling, plane only where it catches, in the wet season, then seal the top edge and the planed edge with paint or varnish. Bare timber edges drink humidity - most doors are sealed on the faces and naked on top.

It hit a fence post last summer, which I had entirely forgotten about.

The arrestor screen was half blocked, though it still would not fire with the muffler off entirely.

Yes. The pop means the nail lost its grip, so the fix is fresh fasteners holding the board, not the same nail hammered back in.

Spot patch, but do it properly and change the paint.

A dozen nail pops is a two hour job: drive a screw about 5cm above and below each pop into the stud, deal with the popped nail, then three thin coats of compound with sanding between, then prime the patches. Spot patches show almost entirely because people do one thick coat feathered over 10cm instead of three thin coats feathered over 30cm.

The bigger lever is sheen. Mid sheen on a wall with history is unforgiving - it is doing the work of showing every repair. A matt or eggshell finish hides ordinary wall texture and will make more difference than a full skim would.

Skimming a whole wall as your first plastering job on a free weekend is how you end up sanding for three days.

Try cleaning it first for nothing. You said gummed, not damaged. Bowl off, jets out, cleaner, air, back together. If that fixes it you have spent an evening and can buy the spare carburettor when it eventually needs one.

Wiggle the shaft side to side before you order anything. If there is play, you have your answer for free.