DACTAL.org
@dactal.org
A notional organization for an experimental query language.
You have a rack with no good Scrabble plays. Which tile is the best to unload?
Here's that question asked in DACTAL.
Here's that question asked in DACTAL.
November 5, 2025 at 12:25 AM
You have a rack with no good Scrabble plays. Which tile is the best to unload?
Here's that question asked in DACTAL.
Here's that question asked in DACTAL.
Structurally tiny but useful addition. Inverting a graph arc can now be done like this:
.inverse name>forward property
E.g., in data where each album has track objects, but the tracks don't refer back to their album:
.album>tracks
adds an "album" property to each track while traversing.
.inverse name>forward property
E.g., in data where each album has track objects, but the tracks don't refer back to their album:
.album>tracks
adds an "album" property to each track while traversing.
November 3, 2025 at 11:24 PM
Structurally tiny but useful addition. Inverting a graph arc can now be done like this:
.inverse name>forward property
E.g., in data where each album has track objects, but the tracks don't refer back to their album:
.album>tracks
adds an "album" property to each track while traversing.
.inverse name>forward property
E.g., in data where each album has track objects, but the tracks don't refer back to their album:
.album>tracks
adds an "album" property to each track while traversing.
Tiny additions. Filtering items that have any, or no, values for a property can now be done with:
:property+
:property-
as shorthand for the subquery forms:
:(.property)
:-(.property)
But the subquery forms are still needed for any path more complicated than a single property.
:property+
:property-
as shorthand for the subquery forms:
:(.property)
:-(.property)
But the subquery forms are still needed for any path more complicated than a single property.
November 3, 2025 at 11:20 PM
Tiny additions. Filtering items that have any, or no, values for a property can now be done with:
:property+
:property-
as shorthand for the subquery forms:
:(.property)
:-(.property)
But the subquery forms are still needed for any path more complicated than a single property.
:property+
:property-
as shorthand for the subquery forms:
:(.property)
:-(.property)
But the subquery forms are still needed for any path more complicated than a single property.
I should pretend I designed this feature intentionally, but in reality I just realized that because
something/someprop.10
gives us the first 10 items from each group, you can also truncate each group to 10 items in place, like this:
something/someprop|of=10
something/someprop.10
gives us the first 10 items from each group, you can also truncate each group to 10 items in place, like this:
something/someprop|of=10
October 8, 2025 at 5:47 PM
I should pretend I designed this feature intentionally, but in reality I just realized that because
something/someprop.10
gives us the first 10 items from each group, you can also truncate each group to 10 items in place, like this:
something/someprop|of=10
something/someprop.10
gives us the first 10 items from each group, you can also truncate each group to 10 items in place, like this:
something/someprop|of=10
The Gallery now includes a DACTAL Sudoku solver! Or, more accurately, a Sudoku solution-process revealer.
dactal.org/dactal.html?...
dactal.org/dactal.html?...
October 2, 2025 at 4:14 PM
The Gallery now includes a DACTAL Sudoku solver! Or, more accurately, a Sudoku solution-process revealer.
dactal.org/dactal.html?...
dactal.org/dactal.html?...
This may not interest you until you suddenly hit a case where you need it, but you can now short-circuit a traverse operation when it gets its first results, like this:
some data.<(.some expensive subquery)
instead of running the subquery N times only to throw away N-1 of the results.
some data.<(.some expensive subquery)
instead of running the subquery N times only to throw away N-1 of the results.
October 1, 2025 at 4:03 PM
This may not interest you until you suddenly hit a case where you need it, but you can now short-circuit a traverse operation when it gets its first results, like this:
some data.<(.some expensive subquery)
instead of running the subquery N times only to throw away N-1 of the results.
some data.<(.some expensive subquery)
instead of running the subquery N times only to throw away N-1 of the results.
DACTAL,ATProto.fun
BlueSky is not just an app, it's an experiment in public data, but the experiment is hidden under a normal-looking app that doesn't really give you different powers than you have elsewhere.
If you're curious to see under this surface, I made a thing to help.
Explained here: furia.com/log/518
If you're curious to see under this surface, I made a thing to help.
Explained here: furia.com/log/518
September 17, 2025 at 4:01 PM
DACTAL,ATProto.fun
A start/labeling operation can append to the existing list with
?label=+(subquery)
??label=+(subquery)
And a start operation can use the value ONLY as a reference with
=label
=type
Thus this form for setting a label if it isn't already set:
??district size=(=district size;~1000)
?label=+(subquery)
??label=+(subquery)
And a start operation can use the value ONLY as a reference with
=label
=type
Thus this form for setting a label if it isn't already set:
??district size=(=district size;~1000)
August 30, 2025 at 2:01 PM
A start/labeling operation can append to the existing list with
?label=+(subquery)
??label=+(subquery)
And a start operation can use the value ONLY as a reference with
=label
=type
Thus this form for setting a label if it isn't already set:
??district size=(=district size;~1000)
?label=+(subquery)
??label=+(subquery)
And a start operation can use the value ONLY as a reference with
=label
=type
Thus this form for setting a label if it isn't already set:
??district size=(=district size;~1000)
Sometimes you want (or just need) the results of another query in the middle of the one you're writing. Highlight the subquery and hit Command-U to unpack it.
August 27, 2025 at 3:08 PM
Sometimes you want (or just need) the results of another query in the middle of the one you're writing. Highlight the subquery and hit Command-U to unpack it.
Saved relative queries let you encapsulate complexity, like
AEIODLT
.possible words
.by added letter
And now you can decapsulate it again with command-I (for Inline).
AEIODLT
.possible words
.by added letter
And now you can decapsulate it again with command-I (for Inline).
August 22, 2025 at 3:36 PM
Saved relative queries let you encapsulate complexity, like
AEIODLT
.possible words
.by added letter
And now you can decapsulate it again with command-I (for Inline).
AEIODLT
.possible words
.by added letter
And now you can decapsulate it again with command-I (for Inline).
Also: added regex filtering. And then, only a little later, added documentation of regex filtering.
August 16, 2025 at 12:52 AM
Also: added regex filtering. And then, only a little later, added documentation of regex filtering.
If anybody can vibe-code embryonic ideas into slick apps, we're gonna get a lot of slick apps sealed airlessly around ideas that never got a chance to become.
I want the opposite of that, unslick explorers of exposed exologic. We only start at the beginning.
dactal.org/Scrabble.html
I want the opposite of that, unslick explorers of exposed exologic. We only start at the beginning.
dactal.org/Scrabble.html
August 15, 2025 at 6:37 PM
If anybody can vibe-code embryonic ideas into slick apps, we're gonna get a lot of slick apps sealed airlessly around ideas that never got a chance to become.
I want the opposite of that, unslick explorers of exposed exologic. We only start at the beginning.
dactal.org/Scrabble.html
I want the opposite of that, unslick explorers of exposed exologic. We only start at the beginning.
dactal.org/Scrabble.html
Yesterday's excellent tiny realization: subtractive traversal is a set-disjunction operation, but subtractive traversal with duplicates is itemwise removal.
...x=(1,1,2,2,3,3),y=(2)
.x,-y → 1,3
..x,y → 1,1,2,3,3
...x=(1,1,2,2,3,3),y=(2)
.x,-y → 1,3
..x,y → 1,1,2,3,3
August 8, 2025 at 1:52 PM
Yesterday's excellent tiny realization: subtractive traversal is a set-disjunction operation, but subtractive traversal with duplicates is itemwise removal.
...x=(1,1,2,2,3,3),y=(2)
.x,-y → 1,3
..x,y → 1,1,2,3,3
...x=(1,1,2,2,3,3),y=(2)
.x,-y → 1,3
..x,y → 1,1,2,3,3
August 1, 2025 at 7:13 PM
On one hand, we already know all the prime numbers below 1000.
On the other, here's the Sieve of Eratosthenes in a DACTAL query.
dactal.org/dactal.html?...
On the other, here's the Sieve of Eratosthenes in a DACTAL query.
dactal.org/dactal.html?...
DACTAL
dactal.org
August 1, 2025 at 4:52 PM
On one hand, we already know all the prime numbers below 1000.
On the other, here's the Sieve of Eratosthenes in a DACTAL query.
dactal.org/dactal.html?...
On the other, here's the Sieve of Eratosthenes in a DACTAL query.
dactal.org/dactal.html?...
I give you your carets back. ^ and ^^ for comments have been replaced by ??? for notes, so ^ is no longer a reserved character.
??? your notes here
Notes in DACTAL are a null operation, not meta-text, which has some cool implications that I haven't thought of yet.
??? your notes here
Notes in DACTAL are a null operation, not meta-text, which has some cool implications that I haven't thought of yet.
July 31, 2025 at 8:43 PM
I give you your carets back. ^ and ^^ for comments have been replaced by ??? for notes, so ^ is no longer a reserved character.
??? your notes here
Notes in DACTAL are a null operation, not meta-text, which has some cool implications that I haven't thought of yet.
??? your notes here
Notes in DACTAL are a null operation, not meta-text, which has some cool implications that I haven't thought of yet.
I moved the inline-math handling to a more sensible place in the resolution process, so now you can use it in traversal, grouping, sorting, or either form of annotation.
July 25, 2025 at 3:41 PM
I moved the inline-math handling to a more sensible place in the resolution process, so now you can use it in traversal, grouping, sorting, or either form of annotation.
Realized and fixed an early-implementation laziness that evaluated :x=y as :(.x:@1:=y) instead of :(.x:=y)
July 23, 2025 at 6:07 PM
Realized and fixed an early-implementation laziness that evaluated :x=y as :(.x:@1:=y) instead of :(.x:=y)
To go with /@x, which groups consecutive sequences with shared x properties, you can now also do /@@y to group sequences that start with each occurrence of y.
E.g grouping streams into listening sessions broken by gaps of more than 10 minutes between songs:
E.g grouping streams into listening sessions broken by gaps of more than 10 minutes between songs:
![listening sample....pairs
||px=@,
clock=(....(.pair.ts),timediff);
this<pair;
px=(this.px);
clock=(this.clock);
silence=[=clock-ms_played];
restart=(.silence:>60000);
.(.pair:px=1,@2)/@@restart](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:4n4l7srkx4g6dqu6oxhpy4lr/bafkreifssrzxsrfuzwh3zcrjhnc2dmtxq4j27aecqvkzfrzzzzkizinvzy@jpeg)
July 18, 2025 at 2:16 AM
To go with /@x, which groups consecutive sequences with shared x properties, you can now also do /@@y to group sequences that start with each occurrence of y.
E.g grouping streams into listening sessions broken by gaps of more than 10 minutes between songs:
E.g grouping streams into listening sessions broken by gaps of more than 10 minutes between songs:
Realized that the recently-added |x@ shortcut, for annotating a list of items with their list indexes, was actually the beginning of the missing feature for doing running ranks/counts/totals.
e.g. cumulative duration:
tracks2||cumulative duration=@@duration
dactal.org/dactal.html#...
e.g. cumulative duration:
tracks2||cumulative duration=@@duration
dactal.org/dactal.html#...
DACTAL
dactal.org
July 10, 2025 at 3:25 PM
Realized that the recently-added |x@ shortcut, for annotating a list of items with their list indexes, was actually the beginning of the missing feature for doing running ranks/counts/totals.
e.g. cumulative duration:
tracks2||cumulative duration=@@duration
dactal.org/dactal.html#...
e.g. cumulative duration:
tracks2||cumulative duration=@@duration
dactal.org/dactal.html#...
Huh, worked out the Wordle logic in DACTAL form.
July 4, 2025 at 9:29 PM
Huh, worked out the Wordle logic in DACTAL form.
The new ?? label operation has the same labeling options as a ? start operation, but relative to, and without changing, the current list:
tracks|others=(??this .artist.tracks:-(this))
This replaces the more limited special-casing of "=_" in starts:
tracks|others=(?this=_ .artist.tracks:-(this))
tracks|others=(??this .artist.tracks:-(this))
This replaces the more limited special-casing of "=_" in starts:
tracks|others=(?this=_ .artist.tracks:-(this))
June 28, 2025 at 3:26 PM
The new ?? label operation has the same labeling options as a ? start operation, but relative to, and without changing, the current list:
tracks|others=(??this .artist.tracks:-(this))
This replaces the more limited special-casing of "=_" in starts:
tracks|others=(?this=_ .artist.tracks:-(this))
tracks|others=(??this .artist.tracks:-(this))
This replaces the more limited special-casing of "=_" in starts:
tracks|others=(?this=_ .artist.tracks:-(this))
Today in obsessive buffing of rough spots:
Filter negation is now done with -, instead of !.
tracks:-=Amaranth
This was the only use of ! as a suboperator, which frees it up to be the repeat operator, instead of ??.
messages._,replies!
Filter negation is now done with -, instead of !.
tracks:-=Amaranth
This was the only use of ! as a suboperator, which frees it up to be the repeat operator, instead of ??.
messages._,replies!
June 19, 2025 at 5:19 PM
Today in obsessive buffing of rough spots:
Filter negation is now done with -, instead of !.
tracks:-=Amaranth
This was the only use of ! as a suboperator, which frees it up to be the repeat operator, instead of ??.
messages._,replies!
Filter negation is now done with -, instead of !.
tracks:-=Amaranth
This was the only use of ! as a suboperator, which frees it up to be the repeat operator, instead of ??.
messages._,replies!
Reposted by DACTAL.org
Today's tiny DACTAL tweak that probably affects nobody: in an aggregation, semicolons explicitly identify aggregators. So this query is a count of counts; the first "count" is a property (of groups), the second "count" is the aggregator.
tracks/artist...count;count
tracks/artist...count;count
June 18, 2025 at 8:49 PM
Today's tiny DACTAL tweak that probably affects nobody: in an aggregation, semicolons explicitly identify aggregators. So this query is a count of counts; the first "count" is a property (of groups), the second "count" is the aggregator.
tracks/artist...count;count
tracks/artist...count;count