05:21 <abu[7]> Hi pablo_escoberg, I check your pastebin

05:21 <abu[7]> So the task (task (: Fd) (prinl (in (: Fd)))) is not right

05:23 <abu[7]> The 'in' has an empty body, so it does not read anything

05:23 <abu[7]> (task (: Fd) (prinl (in (: Fd) (line T))))

05:23 <abu[7]> But you need to check also for EOF

05:24 <abu[7]> Another problem is that when the task runs, it will not

05:25 <abu[7]> it will not be in the context of the object

05:25 <abu[7]> so Fd is no accessible

05:26 <abu[m]> oops

05:29 <abu[m]> So 'Fd' must be encapsulated inside the task's environment

05:30 <abu[m]> Not tested: http://pb1n.de/?e0bc90

05:34 <abu[m]> I changed only the task

05:34 <abu[m]> There are other issues, e.g. (out (: Fd) (pack QueryStr sqlite_postfix)) won't output anything to the file descriptor

05:35 <abu[m]> Better perhaps (out (: Fd) (prinl QueryStr sqlite_postfix))

05:35 <abu[m]> where sqlite_postfix is probably *Sqlite_postfix

14:15 <pablo_escoberg> Awesome. Thanks guys.

14:16 <abu[m]> You are starting with the difficult aspects of pil ;)

14:25 <pablo_escoberg> I also learned to swim by jumping in the deep end :).

14:25 <pablo_escoberg> But I definitely feel better knowing that other people find this stuff hard and I'm not quite as stupid as I feel :D

14:26 <abu[m]> good :)

14:26 <abu[m]> As I said, I have to look up such pipe stuff sometimes too

14:29 <pablo_escoberg> A bit OT, and I hope I'm not ruffling any feathers, but has there been any talk of getting on something better than IRC and a mailing list? I get the appeal, it's very old school and grungy, but it's also unwieldy cumbersome. If the discussion has been had and the decision made, I'm fine with it of course, but if not, I'd like to bring it up...

14:30 <pablo_escoberg> I'm not saying move to reddit or anything like that, but maybe Jabber and Discourse or something?

14:30 <abu[m]> There is also a Matrix group. And we did some conferences for a while via Jitsi (PicCon)

14:31 <abu[m]> #picolisp:7fach.de

14:31 <abu[m]> We could perhaps make video meetings there again

14:31 <abu[m]> (not now, I'm travelling during the next days)

14:31 <pablo_escoberg> My concern is less with the real time stuff and more with the archving

14:32 <pablo_escoberg> searching through mailing list archives and irc logs is a huge pain.

14:32 <abu[m]> This irc channel is well archived

14:32 <abu[m]> https://libera.irclog.whitequark.org/picolisp

14:32 <pablo_escoberg> I've seen the archive, and it seems accessible only by date. Doesn't look like there is any way to search the entire log.

14:33 <pablo_escoberg> unless it's there and I missed it.

14:33 <abu[m]> Hmm, there is a search field, but did not try deeply

14:33 <pablo_escoberg> And the mailing list search works, but it's tough to reply to messages, and there is real overhead in belonging to a mailing list that doesn't exist belonging to a forum.

14:34 <abu[m]> true

14:34 <pablo_escoberg> IAC, if you guys are cool with this setup, I'm cool with it too

14:34 <pablo_escoberg> BUT

14:34 <pablo_escoberg> if it's just like this due to inertia, let's fix it.

14:34 <abu[m]> Not sure what is the best

14:35 <pablo_escoberg> OK, that to me sounds like "not broken"

14:35 <pablo_escoberg> back to coding for me :D

14:35 <abu[m]> Right. There is even a rudimentary "reply" feature in picolisp.com, but not really used

14:36 <pablo_escoberg> Wow. didn't even see it.

14:37 <abu[m]> I feel that perhaps IRC is still the easiest

14:40 <pablo_escoberg> Yeah, if it's not broken, no need to fix it. I was just curious if there was a general desire for improvement, and there does not seem to be.

14:41 <abu[m]> :)

14:46 <pablo_escoberg1> Hmmm... Can you help me understand this:

14:46 <pablo_escoberg1> -> "xxxxx"

14:46 <pablo_escoberg1> : *Run

14:46 <pablo_escoberg1> : (out 4 (prinl "xxxxx"))

14:46 <pablo_escoberg1> -> ((4 (prinl "asdfdsadfasdf")))

14:46 <pablo_escoberg1>

14:46 <pablo_escoberg1> Why isn't the task firing here? Or is it writing out to the Fd?

14:47 <abu[m]> The task wakes up as soon as 4 has data for reading (or is closed)

14:48 <abu[m]> Is 4 open for reading *and* writing? If so, what is on the other side?

14:49 <pablo_escoberg1> it's in the same snipped. 4 is a pipe

14:49 <abu[7]> At least should the task "read off" the data if any are available

14:49 <abu[7]> Otherwise it fires again and again

14:49 <pablo_escoberg1> I just did (in 4 (line)) and got asdfasdfasdf printing over and over. I guess it didn't get flushed, and I guess it is indeed writing to fd 4.

14:50 <pablo_escoberg1> yup, that explains everything. thanks again.

14:50 <abu[7]> (in 4 (line)) is good. Reads off one line

14:52 <pablo_escoberg1> yup, and I'll need to loop and check for EOF. My issue was that I didn't realize it would be writing to the Fd in question rather than stdout. I just need to change it to sth like (out 0 (prinl @)) to see what I'm looking for

14:53 <abu[7]> I use (msg @)

14:53 <abu[7]> best for debug output

14:53 <abu[7]> goes to stderr, and preserves readline() status

14:54 <pablo_escoberg1> Nice! So many cool little syntactic niceties in picolisp I need to learn. You guys really have built something great.

14:54 <pablo_escoberg1> That one will save me tons of pain.

14:59 <pablo_escoberg1> Also, since the thing writes to the affected fd, I assume it reads from it too, so I probably won't need to specify it in the (in (line) part. So it doesn't matter that it's out of scope, I think. I'll play around and let you know.

15:02 <abu[7]> This is what I meant with the most tricky aspect

15:03 <abu[7]> After writing to the Fd, the read might block without a separate task

15:09 <abu[7]> So the behavior of the whole setup depends on how cooperative the process at the other end of the pipe is

15:09 <abu[7]> Especially flushing is important

15:11 <abu[7]> Many programs in Unix don't flush implicitly after end-of-line unless the FD goes to a terminal

15:11 <abu[7]> isatty()

15:12 <abu[7]> But flushig *always* happens if the process terminates and/or closes the pipe

15:12 <abu[7]> That's why single-shot communication is so much easier

15:24 <pablo_escoberg1> oy vey. Looks like I have my work cut out, but I'm thinking I might not need the task after all. Still playing, but need to walk the dog first :).

23:26 <pablo_escoberg1> It works! https://pastebin.com/vSpkdsXe