test please delete

[vetran]

Indeed - and a slight nudge to the bumpers will crack the windscreen too.

If it rains the warranty is void.

[NickFitz]

Changed the bed. Looked up how to configure log rotation in Ubuntu.

Now off to do the shopping

[DoctorStrangelove]

Changed the bed, bedding washed in machine #2 and then hung on the line since it wasn't quite raining at the time.

Cottons done at 90.

Shirts done at 40.

Denim & coloureds done at whatever temperature it decided was a good idea.

All apart from the last now tumble dried.

Yesterday's primer/undercoat given some proper undercoat max voc kind, plus yesterday's primer/undercoat containing cobalt thorium G given another coat.

Just to piss me off even more some delightful cat shat outside the window I was painting and it was so unbelievably disgusting it made me gag.

Solids are bad enough but this was liquid and not in a good way.

More of this year's brambles cut back.

Car loaded up with stuff for the dump.

Off hence here I go.

Living the dream as ever.

I think I preferred the thing that shat outside the back door of the other house over today's offering.

[WTFH]

Been on a conference call for the last hour, during which The Dog (tw) has proceeded to drop some stinkers under my desk. I'm almost in tears here and wondering if the Geneva Convention applies to dogs.

[vetran]

Been on a conference call for the last hour, during which The Dog (tw) has proceeded to drop some stinkers under my desk. I'm almost in tears here and wondering if the Geneva Convention applies to dogs.

Death by Mutt(ard) gas?

[NickFitz]

Shopping done, not that there was that much of it: skipped M&S, and didn't exactly fill my trolley at Sainsbury's.

Got a couple of Not Actually Fried™ southern fried chicken bits from the hot food counter for a late lunch

[DoctorStrangelove]

Dump visited.

A lot quieter during the week.

Then off to Machine Mart.

Almost didn't get there coz the M4 was crawling along at a snail's pace for no readily apparent reason.

Following Machine Mart it was off to Buyology which inhabits part of the old B&Q in Llansamlet.

And after that, off home again.

The drizzle instantly got heavier as soon as I got out of the car.

The final run (denim & colours) had finished in the drier.

After which it was time for the M&S breaded haddock, unfollowed by custard & stewed anything.

Utterly & completely fecked now, so I shall do the dishes & collapse in a heap to watch some telly.

I got the remake of Inglorious Bastards today so maybe that would be a good choice.

[NickFitz]

Well, that was a nice little bug

Twitter tweets have an ID which is a 64 bit integer value, and also include it as a string called id_str, because ten years or so ago when they originally built this stuff, support for 64 bit was a lot less widespread. My stuff, following accepted best practice (that's a bit out of date now, I reckon), has been using the string version, despite Python 3 having support for integers of any size. Most importantly, I use the id_str value in the database.

My stuff for getting newer tweets relies on only asking for tweets since whatever the maximum tweet ID I currently have is. I found that maximum ID using the database MAX function (via Django).

But, sometime in the last few months, tweet IDs rolled over from 18 digits to 19: that is, from 99...99 to 100..00. And, of course, if you ask a database for the maximum value of a string column, it regards a string beginning '9' as greater than one beginning '1'.

So my code was always asking for stuff more recent than a tweet that's now months old, and as it can only get 200 at a time (up to 800 max over multiple calls), it was always making five calls to the API (including the last one that returned nothing, because of the 800 limit), not realising it was getting stuff I already had in the database

Now fixed, using CAST in the database query (via django.db.models.function.Cast) to do the comparison using big integers. Works the database a bit harder, but it's not that busy anyway.

And, as it's just run, a quick look at the log confirms that it seems to be working: 91 new tweets in the last fifteen minutes, rather than chomping through a whole bunch more; and the rate limit hasn't taken a hammering

And with that, I'm off to the Chinese

[mudskipper]

And, of course, if you ask a database for the maximum value of a string column, it regards a string beginning '9' as greater than one beginning '1'.

Doh. Schoolboy error.

[mudskipper]

I had a similar(ish) problem recently in that if you select sum a load of int columns, it assumes the result type must be int, so if there's lots of big numbers you get an overflow error. So you have to convert your ints to bigints before summing. (SQL server, your database may vary)