mmb

It definitely is better than 10.3 (which was just plain broken) but also <10.3. It isn't perfect but as I said in another topic for my project (680K unique lines of code) it is getting more reliable/faster the cleaner I make my source tree. With cleaner I mean things like only use units that are really needed, reducing coupling, smallest scope of identifiers possible (= members not public when protected will do, not protected when private will do; up to a whole class not in the interface section but exposing just an interface and a factory etc etc). Things like these are a good idea anyway so actually the tool "helping" me with the required discipline could be considered a +. 🙂 I am now at the point that error insight and code insight work reliably most of the time and wait times till red wiggles disappear is down to seconds.

It does work but for some reason not on the symbol where you define it itself. In any case I cannot get it reliably to work except in simple test applications. I tried it in both 10.3 and 10.4. The difference being that in the same project the line without the <summary> tag works in 10.4 but not in 10.3.

Same, my gut feeling is that the current implementation is pretty much a happy flow, if it works it is amazing but seemingly simple things can throw it off. I find that the cleaner I make my project tree (only stuff a unit really uses in the uses clauses), untangling unnecessary inheritance trees, removing visual form/datamodule inheritance (which I unfortunately "inherited" from my predecessors) and so on the better code insight/error insight/navigation/inline xml documentation work. TL;DR Put some uncle bob in your project tree, it helps in more ways than one.

Just noticed that in 10.4 /// Lorem Ipsum.... works just as well as /// <summary> Lorem Ipsum.... </summary> This makes in source XML documentation much easier to write. I could not find anything about this in the release notes (or I overlooked). Is this a side effect of the LSP?

I'm porting to various environments, depending on the need of the application. I am almost finished with a port of a Mac application to Swift (using Xcode). Granted this was more of a rewrite but since our customer requires 64 bit support (so they can deploy to the Mac App Store) and refuses to pay us without 64 bit support so I had no choice. Since from a customers perspective the application is exactly the same I cannot charge for porting, especially since I was the one that convinced the customer to go with Delphi when we started this project two years ago.... I feel like an idiot for having defended Emba/Delphi and now I am paying the price... Other tools (mainly data analysis) I've ported to Python. This was also more of a rewrite and required a completely different mindset. My reservations "Python is just a script language" and "not type safe?" turned out to be completely unfounded; Properly written ("Pythonic") code is beautiful! Third party (open source) availability is simply amazing (micro-service architecture using ZeroMQ). I did briefly consider Lazarus but decided against it. I could not afford to take any risks (swift on Mac OS is basically a given w.r.t. support) and Python runs anywhere. Stubbornly staying with object pascal also "felt" wrong; if the tool/language I trusted and relied upon for 20 years betrayed me I might as well take a deep dive and go in a completely different direction. As a side-note: having seen what modern IDE's can do (Xcode/PyCharm) and what a joy they are to work with I highly doubt Emba will be able to catch up even if they start treating their customers as customers again. I'm not ruling out ever taking a subscription again but it is going to take some really impressive releases and flawless ethical behaviour for Emba to regain my trust.

Same. I've let my subscription expire earlier this year mostly due to the MacOS64 support still not being there and the beyond ridiculous "error insight not supporting the new language features". The current drama with the registration limit confirms that I made the right choice. I am in the process of porting all my current customers applications to other environments and this is costing me a small fortune as I cannot in good conscience bill my customers for that. I really hope to be proven wrong but to be honest I have given up hope.

I've let my enterprise subscription expire yesterday. It breaks my heart after two decades of working with Delphi but as many others I just lost all confidence in emba/idera. IMHO complaining but at the same time keep paying ridiculously high subscription fee (compared to other development environments) sends them the wrong message. I'll gladly resubscribe if things turn around or when Delphi eventually gets auctioned off to a better care-taker.