[bug#45299] [PATCH] maint: Require Guile >= 2.2.6.

[bug#45299] [PATCH] maint: Require Guile >= 2.2.6.

[Ludovic Courtès]

* configure.ac: For Guile 2.2, require 2.2.6 or later.
* guix/gexp.scm (define-syntax-parameter-once): Remove.
Use 'define-syntax-parameter' instead.
* guix/mnoads.scm: Likewise.
* guix/inferior.scm (proxy)[select*]: Remove.
* guix/scripts/publish.scm <top level>: Remove replacement for (@@ (web
http) read-header-line).
* guix/store/deduplication.scm (counting-wrapper-port): Remove.
(nar-sha256): Call 'port-position' on PORT to compute SIZE.
---
 configure.ac                 |  2 +-
 guix/gexp.scm                | 15 ++-------------
 guix/inferior.scm            | 11 +----------
 guix/monads.scm              | 15 ++-------------
 guix/scripts/publish.scm     | 26 --------------------------
 guix/store/deduplication.scm | 32 ++++----------------------------
 6 files changed, 10 insertions(+), 91 deletions(-)

diff --git a/configure.ac b/configure.ac
index a5bdf24e93..afb449950f 100644
--- a/configure.ac
+++ b/configure.ac
@@ -102,7 +102,7 @@ if test "x$GUILD" = "x"; then
 fi
 
 if test "x$GUILE_EFFECTIVE_VERSION" = "x2.2"; then
-  PKG_CHECK_MODULES([GUILE], [guile-2.2 >= 2.2.3])
+  PKG_CHECK_MODULES([GUILE], [guile-2.2 >= 2.2.6])
 fi
 
 dnl Get CFLAGS and LDFLAGS for libguile.
diff --git a/guix/gexp.scm b/guix/gexp.scm
index 051831238e..764c89a187 100644
--- a/guix/gexp.scm
+++ b/guix/gexp.scm
@@ -1317,18 +1317,7 @@ and in the current monad setting (system type, etc.)"
                    reference->sexp (gexp-references exp))))
     (return (apply (gexp-proc exp) args))))
 
-(define-syntax-rule (define-syntax-parameter-once name proc)
-  ;; Like 'define-syntax-parameter' but ensure the top-level binding for NAME
-  ;; does not get redefined.  This works around a race condition in a
-  ;; multi-threaded context with Guile <= 2.2.4: <https://bugs.gnu.org/27476>.
-  (eval-when (load eval expand compile)
-    (define name
-      (if (module-locally-bound? (current-module) 'name)
-          (module-ref (current-module) 'name)
-          (make-syntax-transformer 'name 'syntax-parameter
-                                   (list proc))))))
-
-(define-syntax-parameter-once current-imported-modules
+(define-syntax-parameter current-imported-modules
   ;; Current list of imported modules.
   (identifier-syntax '()))
 
@@ -1339,7 +1328,7 @@ environment."
                          (identifier-syntax modules)))
     body ...))
 
-(define-syntax-parameter-once current-imported-extensions
+(define-syntax-parameter current-imported-extensions
   ;; Current list of extensions.
   (identifier-syntax '()))
 
diff --git a/guix/inferior.scm b/guix/inferior.scm
index 77820872b3..2fe91beaab 100644
--- a/guix/inferior.scm
+++ b/guix/inferior.scm
@@ -469,22 +469,13 @@ is similar to the sexp returned by 'package-provenance' for regular packages."
   "Proxy communication between CLIENT and BACKEND until CLIENT closes the
 connection, at which point CLIENT is closed (both CLIENT and BACKEND must be
 input/output ports.)"
-  (define (select* read write except)
-    ;; This is a workaround for <https://bugs.gnu.org/30365> in Guile < 2.2.4:
-    ;; since 'select' sometimes returns non-empty sets for no good reason,
-    ;; call 'select' a second time with a zero timeout to filter out incorrect
-    ;; replies.
-    (match (select read write except)
-      ((read write except)
-       (select read write except 0))))
-
   ;; Use buffered ports so that 'get-bytevector-some' returns up to the
   ;; whole buffer like read(2) would--see <https://bugs.gnu.org/30066>.
   (setvbuf client 'block 65536)
   (setvbuf backend 'block 65536)
 
   (let loop ()
-    (match (select* (list client backend) '() '())
+    (match (select (list client backend) '() '())
       ((reads () ())
        (when (memq client reads)
          (match (get-bytevector-some client)
diff --git a/guix/monads.scm b/guix/monads.scm
index 6924471345..6ae616aca9 100644
--- a/guix/monads.scm
+++ b/guix/monads.scm
@@ -274,23 +274,12 @@ more optimizations."
                    (_
                     #'generic-name))))))))))
 
-(define-syntax-rule (define-syntax-parameter-once name proc)
-  ;; Like 'define-syntax-parameter' but ensure the top-level binding for NAME
-  ;; does not get redefined.  This works around a race condition in a
-  ;; multi-threaded context with Guile <= 2.2.4: <https://bugs.gnu.org/27476>.
-  (eval-when (load eval expand compile)
-    (define name
-      (if (module-locally-bound? (current-module) 'name)
-          (module-ref (current-module) 'name)
-          (make-syntax-transformer 'name 'syntax-parameter
-                                   (list proc))))))
-
-(define-syntax-parameter-once >>=
+(define-syntax-parameter >>=
   ;; The name 'bind' is already taken, so we choose this (obscure) symbol.
   (lambda (s)
     (syntax-violation '>>= ">>= (bind) used outside of 'with-monad'" s)))
 
-(define-syntax-parameter-once return
+(define-syntax-parameter return
   (lambda (s)
     (syntax-violation 'return "return used outside of 'with-monad'" s)))
 
diff --git a/guix/scripts/publish.scm b/guix/scripts/publish.scm
index c31cef3181..5a865c838d 100644
--- a/guix/scripts/publish.scm
+++ b/guix/scripts/publish.scm
@@ -824,32 +824,6 @@ example: \"/foo/bar\" yields '(\"foo\" \"bar\")."
 (define %http-write
   (@@ (web server http) http-write))
 
-(match (list (major-version) (minor-version) (micro-version))
-  (("2" "2" "5")                                  ;Guile 2.2.5
-   (let ()
-     (define %read-line (@ (ice-9 rdelim) %read-line))
-     (define bad-header (@@ (web http) bad-header))
-
-     ;; XXX: Work around <https://bugs.gnu.org/36350> by reverting to the
-     ;; definition of 'read-header-line' as found in 2.2.4 and earlier.
-     (define (read-header-line port)
-       "Read an HTTP header line and return it without its final CRLF or LF.
-Raise a 'bad-header' exception if the line does not end in CRLF or LF,
-or if EOF is reached."
-       (match (%read-line port)
-         (((? string? line) . #\newline)
-          ;; '%read-line' does not consider #\return a delimiter; so if it's
-          ;; there, remove it.  We are more tolerant than the RFC in that we
-          ;; tolerate LF-only endings.
-          (if (string-suffix? "\r" line)
-              (string-drop-right line 1)
-              line))
-         ((line . _)                              ;EOF or missing delimiter
-          (bad-header 'read-header-line line))))
-
-     (set! (@@ (web http) read-header-line) read-header-line)))
-  (_ #t))
-
 (define (strip-headers response)
   "Return RESPONSE's headers minus 'Content-Length' and our internal headers."
   (fold alist-delete
diff --git a/guix/store/deduplication.scm b/guix/store/deduplication.scm
index a72a43bf79..cd9660174c 100644
--- a/guix/store/deduplication.scm
+++ b/guix/store/deduplication.scm
@@ -37,38 +37,14 @@
             dump-file/deduplicate
             copy-file/deduplicate))
 
-;; XXX: This port is used as a workaround on Guile <= 2.2.4 where
-;; 'port-position' throws to 'out-of-range' when the offset is great than or
-;; equal to 2^32: <https://bugs.gnu.org/32161>.
-(define (counting-wrapper-port output-port)
-  "Return two values: an output port that wraps OUTPUT-PORT, and a thunk to
-retrieve the number of bytes written to OUTPUT-PORT."
-  (let ((byte-count 0))
-    (values (make-custom-binary-output-port "counting-wrapper"
-                                            (lambda (bytes offset count)
-                                              (put-bytevector output-port bytes
-                                                              offset count)
-                                              (set! byte-count
-                                                (+ byte-count count))
-                                              count)
-                                            (lambda ()
-                                              byte-count)
-                                            #f
-                                            (lambda ()
-                                              (close-port output-port)))
-            (lambda ()
-              byte-count))))
-
 (define (nar-sha256 file)
   "Gives the sha256 hash of a file and the size of the file in nar form."
-  (let*-values (((port get-hash) (open-sha256-port))
-                ((wrapper get-size) (counting-wrapper-port port)))
-    (write-file file wrapper)
-    (force-output wrapper)
+  (let-values (((port get-hash) (open-sha256-port)))
+    (write-file file port)
     (force-output port)
     (let ((hash (get-hash))
-          (size (get-size)))
-      (close-port wrapper)
+          (size (port-position port)))
+      (close-port port)
       (values hash size))))
 
 (define (tempname-in directory)
-- 
2.29.2

[bug#45299] [PATCH] maint: Require Guile >= 2.2.6.

[Efraim Flashner]

[-- Attachment #1: Type: text/plain, Size: 920 bytes --]

Debian 10 (the current release) has guile-2.2.4

(ins)efraim@E5400 ~$ ssh do1 apt-cache policy guile-2.2
guile-2.2:
  Installed: (none)
  Candidate: 2.2.4+1-2+deb10u1
  Version table:
     2.2.7+1-5.4 200
        200 http://deb.debian.org/debian sid/main amd64 Packages
     2.2.4+1-2+deb10u1 500
        500 http://mirrors.digitalocean.com/debian buster/main amd64 Packages
        500 http://deb.debian.org/debian buster/main amd64 Packages

guile-3 is currently in unstable

(ins)efraim@E5400 ~$ ssh do1 apt-cache policy guile-3.0
guile-3.0:
  Installed: (none)
  Candidate: 3.0.4-3
  Version table:
     3.0.4-3 200
        200 http://deb.debian.org/debian sid/main amd64 Packages

-- 
Efraim Flashner   <efraim@flashner.co.il>   אפרים פלשנר
GPG key = A28B F40C 3E55 1372 662D  14F7 41AA E7DC CA3D 8351
Confidentiality cannot be guaranteed on emails sent or received unencrypted

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

[bug#45299] [PATCH] maint: Require Guile >= 2.2.6.

[Ludovic Courtès]

Hi Efraim,

Efraim Flashner <efraim@flashner.co.il> skribis:

> Debian 10 (the current release) has guile-2.2.4

Hmm OK.  Should it hold us back, though?

2.2.6 was released in June 2019.  Also, I expect most users to either
use Vagrant’s brand new ‘guix’ package or to use our binary tarball.

Thoughts?

Ludo’.

[bug#45299] [PATCH] maint: Require Guile >= 2.2.6.

[Vagrant Cascadian]

[-- Attachment #1: Type: text/plain, Size: 1153 bytes --]

On 2020-12-18, Ludovic Courtès wrote:
> Efraim Flashner <efraim@flashner.co.il> skribis:
>
>> Debian 10 (the current release) has guile-2.2.4

> Hmm OK.  Should it hold us back, though?

Debian 10 doesn't have many of the guile-* build dependencies, so I
would not recommend the current Debian stable release block bumping the
minimum guile 2.2 version for guix...


> 2.2.6 was released in June 2019.

Debian bullseye (the testing branch) currently has guile-2.2 at 2.2.7:

  https://tracker.debian.org/pkg/guile-2.2

Debian bullseye is likely to become Debian stable mid-2021, starting to
freeze in early 2021.

It should have all the dependencies for building guix against guile-2.2,
and *almost* all for building guix against guile-3.0...


> Also, I expect most users to either use Vagrant’s brand new ‘guix’
> package or to use our binary tarball.

The guix package may not make it out of Debian experimental any time
soon... unless I switch it to use guile-2.2 (and track down the
corresponding test suite failures) or the guile-gnutls test suite issues
with guile-3.0 get fixed.


live well,
  vagrant

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 227 bytes --]

bug#45299: [PATCH] maint: Require Guile >= 2.2.6.

[Ludovic Courtès]

Hi,

Vagrant Cascadian <vagrant@debian.org> skribis:

> On 2020-12-18, Ludovic Courtès wrote:
>> Efraim Flashner <efraim@flashner.co.il> skribis:
>>
>>> Debian 10 (the current release) has guile-2.2.4
>
>> Hmm OK.  Should it hold us back, though?
>
> Debian 10 doesn't have many of the guile-* build dependencies, so I
> would not recommend the current Debian stable release block bumping the
> minimum guile 2.2 version for guix...

Alright.  Pushed as 4f621a2b003e85d480999e4d0630e9dc3de85bc3!

> The guix package may not make it out of Debian experimental any time
> soon... unless I switch it to use guile-2.2 (and track down the
> corresponding test suite failures) or the guile-gnutls test suite issues
> with guile-3.0 get fixed.

Noted, I should keep investigating.

Thanks for your feedback!

Ludo’.