Software, Outlook, BGP, Nethack, CF WAF, Ubuntu, Mixcloud
- Development vs Maintenance? With some projects, it feels like it's hard to make development progress. Because the project maintenance is starting to take such an heavy toll. Project dependencies, interfaces, everything is being changed all the time. So just to keep the project working requires a lot of work. Which means that there's hardly any new development. I guess this is very common with large projects. I'll have to rewrite the code to use FTSModel with SearchFields and so on with Peewee 3.X. Anyway, that's something which was on my list already, because the old way of indexing that data was inefficient, I mean not indexing, but storing, because the indexed data was basically stored two or three times. I'll check latest JSON extensions too, and maybe I'll get a solution where data is stored only once, which would be preferable unless it delivers too high search / retrieval performance hit. - I'm just complaining because I'm busy. Charles is doing great work with Peewee ORM . If it wouldn't be great work, I wouldn't be using it to begin with.
- Outlook performance varies wildly. I don't know if they're connecting clients to random data centers or something. Sometimes everything is very fast, and sometimes so slow most of operations simply timeout. Something isn't working in optimal way now. Same problem is present with both protocols: ActiveSync & IMAP.
- A nice post by Cloudflare how BGP leak was used to poison DNS and attack users and steal cryptocurrencies.
- Nethack 3.6.1 released - Some things are just sticking around. Which is quite awesome nowadays. So much code gets scrapped without ever seeing release. This reminded me about one project which was dead for two years and have been now resurrected quite quickly by updating all libraries, relations, message formats, compiler compatibility etc. But it was worth of it.
- Had hard time figuring out one issue where client connecting server was getting status: 400 Bad Request. To summarize the problem, it worked using cURL, but didn't work with the client. But there was one small thing which did lead me stray when debugging. It was the Cloudflare WAF. I didn't know that there was a rule that certain URL paths are only allowed when request Content-Type = application/json and if that's missing, then you'll get 400. I did check everything due that fail. Only when I realized that the missing content type leads to bad request, only with that path, then it was obvious that the header was missing. I tried absolutely everything, special characters in URL, different URL snippets, longer and shorter URLs. At least I did straddle efficiently with excel sheet listing affecting factors. Which did lead me to detecting this error with specific paths, which did lead me figuring out that the client didn't set all headers correctly. After all I figured out that the request object also got method header_items() which helped me to figure out just before making the request what the active headers exactly are.
- Ubuntu 18.04 LTS (Bionic Beaver) - I'll wait over the summer, before installing it. New versions got so many problems, that I don't want to experience that again, especially with systems I do care about. Of course I can run the latest version in VM or using burner laptop where losing everything doesn't matter at all. And wipe & re-install is just normal regular scheduled event. I'm not usually bothering with other than LTS versions nowadays.
- About bad software? Mixcloud next on play queue display is totally broken. It doesn't remove played entries, it shows those still on next / upcoming list. Also the list doesn't track the tracks when being played. These kind of broken things sometimes annoy me a lot. Btw, this issue has existed at least six months. Is it really too hard to playlist.pop(0), seems to be so. Or just deque.popleft(). Of course it works that you'll manually remove items from the list, but ugh. That's ugly. Also browsing the next list totally sucks, it has sucked all the time. It's slow painful, you don't know how much stuff is in queue and so on. Maybe I should use the listen later feature, instead of play queue.
- I'm wondering if Google+ is showing me spam content on purpose? Why, I'm reliably flagging it. So they might be showing me spam, just to get confirmation if it's spam. Annoying. Is this what AI leads to? Of course it does sense, to use "reliably users" to confirm is something is spam. That's what I would also do to boost spam flagging confidence score. But as user, I fell it's extremely annoying. Very Nth post is pretty much guaranteed to be spam. I think I'll stop flagging it in future.