Hey, guys! Does anyone know if Racket has something like Factor's with-directory?
This is what I wrote:
(define (with-current-directory dir func)
(define prev-dir (current-directory))
(current-directory dir)
(define result (func))
(current-directory prev-dir)
result)It works, but 1) it feels awkward to define two things for such a simple job (maybe it's just Factor in me talking), and
2) if anyone knows how to make sure the current-directory would be restored in case of exceptions in the func, I would appreciate a hint.
master-db7911c017 on Jan 26 still hangs after a short session, so I have to explore other options.)
(current-directory) is a parameter. So if I remember correctly, you can use parameterize. It's taking care of exceptions. https://docs.racket-lang.org/guide/parameterize.html
1 reply
The only issue is, on Windows its work sessions can be very short and one never knows if there will be a chance to finish what one started.
Hey all, I started on a protobuf lib a while ago - I kind of got distracted but I got to a decent point I feel like I should get it properly working and not just conceptually working. I have the foundations of the thing in pure factor (without any c++ shenanigans and creating objects dynamically instead of having a code generation step like most libs do).
Some technical questions
- This is probably my nervousness at not being a dynamic language person but I want to make sure: is my use of
with-compilation-unitok here.
a. if so, is there a catch to using this during unit-tests? Tests keep failing because words I expect to get defined as they run are throwing errors as not defined - I tried looking at pre-existing tests with this word in it but they don't seem to do anything special? - Is there a way for me to alter the "offset" value of a tuple slot and if so is that something I should avoid? In docs all I can find is that it is an "implementation detail", is it actually just the number of elements into an array of slots (that I assume a tuple is built around)
Some general factor questions:
- my listener history doesn't persist across sessions. I suspect this is implemented to some degree as I have a
.factor_historyin my home directory but it only has history up to an oldsave-imagecall. I can't seem to find docs/code relevant to this - is there some symbol I need to define at startup to read in that file to the listener history recall? - I always forget whether I need bi* or bi@ or 2bi. Has there been a more symbolic notation that is a bit more intuitive in factor? I imagine something like --| for bi@ (2 inputs to one quote), -|-| for 2bi (each input to each quote) and -|| for bi* (on input to both quotes).
you can redefine tuples, the machinery handles it for you if you do, for example:
TUPLE: foo name ;
TUPLE: foo name address ;it will update all tuple instances to add the address slot. reordering works as expected. when you remove/rename a slot they are removed from instances and set to a default initial value, i believe.
@inivekin i'm excited about your protobuf code! if you want a code review, let us know.
1) .factor_history comes from using libreadline and the command-line factor. we currently don't persist listener sessions history, but it wouldn't be that hard to add. so if you do $ rlwrap ./factor then that will participate in "history saving".
2) we've looked at symbolic versions of those and they seem even harder to read than bi/bi*/bi@... one solution is to use cleave/spread/apply with quotations... little bit more verbose, much easier to understand code data flows...
@alex-ilin i wish we could figure out your windows situation, thats not a bug we have been able to reproduce for other windows users. i wonder if it has to do with video graphics drivers? we had some issues on linux with certain buggy GL drivers. are you sure you have the latest for your hardware?
I currently have 3 Lenovo laptops, all based on self-updating Win10x64, each with different graphics cards, and I didn't do anything specific to the video drivers. The latest laptop is ThinkPad L15 Gen 2, with just Intel's integrated graphics. The other two have both integrated and discrete video cards (I believe the integrated one is always used for desktop apps, but I'm not sure whether Factor qualifies as a regular desktop app, or a custom game interface).
First of all, I've got Programmer Dvorak driver v1.2.8 downloaded from here: https://www.kaufmann.no/roland/dvorak/
Second, I have the Russian layout to keep hackers away. Here's my Win-Space switcher screenshot:
But I haven't seen any other apps having such issues, so it should be fixable in Factor in any case.
8c69252. I wonder if I could time-travel back to that state and run the old Factor version in parallel. Of course, some of my vocabs that are updated to the latest version would throw a hissy fit, but for basic stuff it should be possible to run just to prove a point. I'll look for an old enough zip-file on the build server.
https://downloads.factorcode.org/releases/ have the same date-time? I hoped to search for a version based on the date-time of that commit I mentioned.
I'm pretty confident that whatever the problem is, v0.97 doesn't have it.
This was done on the same Win10 laptop, with the Dvorak driver installed.
- I have solved some university homework in Factor, and as part of the assignment I had to write instructions to the teaching assistants who'll check the results about how to run the script. Obviously, I could not tell them to use the bleeding edge version in case something gets broken, I told them to use the latest available release. This means that I had to adapt my solution from the latest version to the older one;
- I have reproduced the hanging issue in v0.98, and I think it's safe to say that v0.97 doesn't have it. Now, the gap between the two is about 4 years IIRC, and there are no binaries in between, which I could run to bisect the history.