- alexandria-devel - mailman3.common-lisp.net

Review MR 27, please
by Robert Goldman 02 May '23

02 May '23

If anyone is still on this mailing list, would you please review this MR and either make requests for fixing it, or merge it? Note that the regression tests are broken on the master branch. One of the reasons I would like to get this MR merged is that I would very much like to finish fixing the regression tests so that we have a secure foundation for any further bug fixes. Thanks, R

3 4

[PATCH] Unnecessary copies and pitfalls in read-stream-content-into-byte-vector
by Sebastian Melzer 14 Feb '23

14 Feb '23

If a byte stream s has exactly n bytes remaining, then (read-stream-content-into-byte-vector s :initial-size n) will result in two unnecessary copies and a total allocation of 4n bytes: - n for the initial buffer; - 2n when the buffer is resized and copied since r-s-c-i-b-v doesn't yet realize that EOF has been reached; - n when the now oversized buffer is downsized and copied at the end. Perversely, this means that a size overestimate is more efficient than a correct estimate. I've attached a possible patch for this below. Sadly there is no peek-byte, so I had to opt for a rather awkward read-byte construction; maybe someone knows a better approach. The patch additionally gets rid of the pitfall where idioms similar to (r-s-c-i-b-v s :initial-size (file-length s)) fail when (file-length s) is zero, but I don't know if broadening the type of initial-size would be considered a breaking change. The internal %length argument also becomes basically unnecessary, but I'd bet someone out there uses it as a workaround. diff --git a/alexandria-1/io.lisp b/alexandria-1/io.lisp --- a/alexandria-1/io.lisp +++ b/alexandria-1/io.lisp @@ -95,29 +95,35 @@ unless it's NIL, which means the system default." (defun read-stream-content-into-byte-vector (stream &key ((%length length)) (initial-size 4096)) "Return \"content\" of STREAM as freshly allocated (unsigned-byte 8) vector." - (check-type length (or null non-negative-integer)) - (check-type initial-size positive-integer) - (do ((buffer (make-array (or length initial-size) - :element-type '(unsigned-byte 8))) - (offset 0) - (offset-wanted 0)) - ((or (/= offset-wanted offset) - (and length (>= offset length))) - (if (= offset (length buffer)) - buffer - (subseq buffer 0 offset))) - (unless (zerop offset) - (let ((new-buffer (make-array (* 2 (length buffer)) - :element-type '(unsigned-byte 8)))) - (replace new-buffer buffer) - (setf buffer new-buffer))) - (setf offset-wanted (length buffer) - offset (read-sequence buffer stream :start offset)))) + (check-type length (or null non-negative-integer)) ; for compatibility + (check-type initial-size non-negative-integer) + (setf initial-size (or length initial-size)) + (let ((result (make-array initial-size :element-type '(unsigned-byte 8))) + (bytes-read 0)) + (loop + (setf bytes-read (read-sequence result stream :start bytes-read)) + (when (and length (>= bytes-read length)) + (return)) + ;; There is no PEEK-BYTE, so we just try to read a byte. + (let ((next-byte (read-byte stream nil nil))) + (when (null next-byte) + (return)) + (let ((new-result (make-array (if (zerop (length result)) + 4096 + (* 2 (length result))) + :element-type '(unsigned-byte 8)))) + (replace new-result result :end1 bytes-read :end2 bytes-read) + (setf (aref new-result bytes-read) next-byte + result new-result) + (incf bytes-read)))) + (if (= bytes-read (length result)) + result + (subseq result 0 bytes-read)))) (defun read-file-into-byte-vector (pathname) "Read PATHNAME into a freshly allocated (unsigned-byte 8) vector." (with-input-from-file (stream pathname :element-type '(unsigned-byte 8)) - (read-stream-content-into-byte-vector stream '%length (file-length stream)))) + (read-stream-content-into-byte-vector stream :initial-size (file-length stream)))) (defun write-byte-vector-into-file (bytes pathname &key (if-exists :error) if-does-not-exist)

2 1

[PATCH] alexandria-1/io.lisp: (read-stream-content-into-string): fix recent regression
by Madhu 22 Aug '22

22 Aug '22

* alexandria-1/io.lisp: (read-stream-content-into-string): fix recent lossage: This addresses problems in commits (13b1575, 6ff6820, 9c97e6f, on 2021-10-24), which changed the element-type of the array used by READ-SEQUENCE to read from the stream from CHARACTER to the STREAM-ELEMENT-TYPE of the stream. On CCL, ECL etc. this breaks nyxt's source/start.lisp:(listen-socket) which basically does (iolib:with-open-socket (s :address-family :local :connect :passive :local-filename "/tmp/foo") (alexandria:read-stream-content-into-string (iolib:accept-connection s))) Iolib returns an octet stream for which the STREAM-ELEMENT-TYPE is not CHARACTER. READ-STREAM-CONTENT-INTO-STRING uses WITH-OUTPUT-TO-STRING and that calls MAKE-STRING-OUTPUT-STREAM, the argument :ELEMENT-TYPE ((UNSIGNED-BYTE 8)) which fails because it must be a subtype of character The present fix is to check if it is not a character stream and force set the element-type to 'character if it is not. --- alexandria-1/io.lisp | 2 ++ 1 file changed, 2 insertions(+) diff --git a/alexandria-1/io.lisp b/alexandria-1/io.lisp index d3be536..8c93dd6 100644 --- a/alexandria-1/io.lisp +++ b/alexandria-1/io.lisp @@ -65,6 +65,8 @@ (defun read-stream-content-into-string (stream &key (buffer-size 4096)) (check-type buffer-size positive-integer) (let ((*print-pretty* nil) (element-type (stream-element-type stream))) + (unless (subtypep element-type 'character) + (setq element-type 'character)) (with-output-to-string (datum nil :element-type element-type) (let ((buffer (make-array buffer-size :element-type element-type))) (loop -- 2.35.1.dirty

3 4

docstrings.lisp
by Philipp Marek 25 Mar '20

25 Mar '20

I noticed that our copy of docstrings.lisp has diverged from SBCL's over the years. Does anyone have a qualified opinion which is "better"? As SBCL is hardcoded in the Makefile anyway why not just remove our own copy and rely on SBCL's version?

1 0

proper-list-length
by Douglas Katzman 13 Mar '20

13 Mar '20

Hi, we have some continuous build tests that compile everything in safety 1 and safety 0. (Yeah, living on the edge, I know.) The latter fails the self-test for proper-list-length.2 If proper-list-length is called with an improper list of at least 1 cons, it signals an error. This is due to the explicit safety declaration in SAFE-ENDP. But if called on an atom, then memory corruption happens because SLOW is initialized sans type checks by (cons (car list) (cdr list)) What would be the intent in that case? To say to the user: "You're holding it wrong"? Thanks, Doug

1 0

Faster implementation of alexandria:median
by Anish Moorthy 27 Nov '19

27 Nov '19

Hello! I wanted to submit a re-implementation of alexandria:median. Currently the function works by fully sorting the sequence, which has a complexity of O(n log n). In the attached patch I provide an implementation as a special case of the quick select algorithm, which brings its complexity down to O(n). The new implementation is more complicated, though I think it's still simpler than some other algorithms in alexandria. Anyways, the performance gains are dramatic enough that I thought I'd at least submit the patch. I did some testing via the timing code in median-timing.lisp, and observed speedups of 5-10x even on sequences with just 100 elements. I could also provide the quick select algorithm as its own function if there's any interest, though I'm not sure what the appetite for adding new features is. - Anish Moorthy

1 0

Can't clone the git repo
by mcdonohe 16 Aug '19

16 Aug '19

Hi, I'm having trouble cloning the git repository. Git tells me the connection is refused. Has anyone else experienced this problem? What should I do? VIVIT

2 1

Please include NOTF
by phoe＠teknik.io 13 Feb '18

13 Feb '18

After brief discussion on #lisp, I hereby suggest a macro for inclusion in Alexandria. The macro in question is NOTF. ## Reason: An example code (from my own project) for negating the value of a place is: (setf (left-now-p (walk-parent shape)) (not (left-now-p (walk-parent shape)))) This code is problematic because a) it causes double evaluation of place subforms, b) it's long and repetitive. I would rather write: (notf (left-now-p (walk-parent shape))) Lisp has modify macros like INCF/DECF for numbers but does not have such a macro for booleans. It seems like a trivial utility, but also one that might be easily overlooked. ## Upsides: The place is evaluated only once. The resulting code is shorter. ## Downsides None that I can see at the moment. ## Code: ;; Please forgive me - I am not used to making PRs by mail. (export 'notf :alexandria) (define-modify-macro notf () not "Sets the value of PLACE to its logical negation and returns the new value.") ---------------- BR Michal "phoe" Herda

1 0

[PATCH] Make MAP-PERMUTATIONS return the original sequence like MAP-(COMBINATIONS|DERANGEMENTS)
by Steve Losh 10 Feb '18

10 Feb '18

MAP-COMBINATIONS and MAP-DERANGEMENTS both return the original sequence. This changes MAP-PERMUTATIONS to match their behavior. MAP-PERMUTATIONS previously returned one of: the original sequence, NIL, or the result of calling the function, depending on which code path happened to be taken. Patch attached. -- Steve Losh

1 0

[alexandria-devel] latest alexandria doesn't build on ECL due to non-standard LOOP
by Anton Vodonosov 18 Jan '18

18 Jan '18

Hello. Alexandria in the latest quicklisp has a regression comparing to quicklisp 2012-02-08. It doesn't build on ECL because on non-standard LOOP construct (it is tolerated by most Lisps, but not by ECL; see here: http://comments.gmane.org/gmane.lisp.ecl.general/3344) The compiler complains about copy-stream method in io.lisp. Build logs: https://cl-test-grid.appspot.com/blob?key=AMIfv97xRRWJgLOa5TvS78VriRK9mQ82m… Best regards, - Anton

2 1