gitup

aafbsd

New Member

Reaction score: 4
Messages: 15

Another reason why regex or at least wildcard matching for ignore would be useful.
 

covacat

Daemon

Reaction score: 801
Messages: 1,465

here is a patch that adds regex support
also adds -I ignore on command line (works like ignore in config file)
a string preceded with a percent sign (%) is considered a regex otherwise a path
i had to modify prune_tree to check for ignores, otherwise you had to ignore work work/port work/port/file to protect a modified file
the only test i've done is
./gitup -v1 -I "%gitup/work/.*/gitup.c$" ports

Diff:
--- work/gitup-0.96/gitup.c    2021-09-05 19:58:01.000000000 +0300
+++ /home/titus/builds/gitup.c    2021-12-01 12:38:45.404988000 +0200
@@ -52,6 +52,7 @@
 #include <string.h>
 #include <unistd.h>
 #include <zlib.h>
+#include <regex.h>
 
 #define GITUP_VERSION     "0.96"
 #define BUFFER_UNIT_SMALL  4096
@@ -122,6 +123,7 @@
     char                *remote_data_file;
     char                *remote_history_file;
     char               **ignore;
+    regex_t           **reg_ignore;                 
     int                  ignores;
     bool                 keep_pack_file;
     bool                 use_pack_file;
@@ -381,11 +383,18 @@
             return (false);
 
     for (x = 0; x < session->ignores; x++) {
+        if(!session->reg_ignore[x]) {
           ignore = session->ignore[x];
                
          if (strncmp(path, ignore, strlen(ignore)) == 0)
                return (true);
+         } else {
+           if(!regexec(session->reg_ignore[x],path,0,NULL,0)) {
+               printf("skipped %s with %s\n",path,session->ignore[x]);
+               return (true);
                }
+          }     
+    }
 
     return (false);
 }
@@ -511,6 +520,8 @@
 
             prune_tree(session, full_path);
         } else {
+       
+          if(!ignore_file(session,full_path,IGNORE_DELETE))
             if (remove(full_path) == -1)
                 err(EXIT_FAILURE,
                     "prune_tree: cannot remove %s",
@@ -3073,6 +3084,50 @@
 }
 
 
+static void
+add_ignore(connector *session,const char *string)
+{
+    char                temp[BUFFER_UNIT_SMALL];
+    int    ret_temp, length;
+    regex_t reg_temp;
+
+    session->ignore = (char **)realloc(
+        session->ignore,
+        (session->ignores + 1)*sizeof(char *));
+
+    session->reg_ignore = (regex_t **)realloc(
+        session->reg_ignore,
+        (session->ignores + 1)*sizeof(regex_t *));
+    if (session->ignore == NULL || session->reg_ignore == NULL)
+        err(EXIT_FAILURE,
+            "load_config_section: malloc");
+    session->reg_ignore[session->ignores] = NULL;       
+    snprintf(temp, sizeof(temp),
+        "%s",
+        string);
+
+    if (temp[0] != '/')
+        snprintf(temp, sizeof(temp),
+            "%s/%s",
+            session->path_target,
+            string);
+    if(string[0] == '%') {
+      ret_temp = regcomp(&reg_temp,string + 1,REG_EXTENDED);
+      if(ret_temp) {
+       warnx("warning: can't compile %s, ignoring\n",string + 1);
+       } else {
+       session->reg_ignore[session->ignores] = malloc(sizeof(regex_t));
+       if(!session->reg_ignore[session->ignores]) {
+       err(EXIT_FAILURE,
+            "load_config_section: malloc2");
+        }
+       memcpy(session->reg_ignore[session->ignores],&reg_temp,sizeof(regex_t));
+       }
+     }
+    length = session->ignores++;
+    session->ignore[length] = strdup(temp);
+       
+}
 /*
  * load_config_section
  *
@@ -3088,7 +3143,6 @@
     char                temp[BUFFER_UNIT_SMALL];
     int                 integer, length = 0;
     bool                boolean, target, path, ignores;
-
     its = ucl_object_iterate_new(section);
 
     while ((pair = ucl_object_iterate_safe(its, true))) {
@@ -3154,27 +3208,7 @@
 
             while ((ignore = ucl_object_iterate_safe(iti, true))) {
                 string = ucl_object_tostring(ignore);
-
-                session->ignore = (char **)realloc(
-                    session->ignore,
-                    (session->ignores + 1)*sizeof(char *));
-
-                if (session->ignore == NULL)
-                    err(EXIT_FAILURE,
-                        "load_config_section: malloc");
-
-                snprintf(temp, sizeof(temp),
-                    "%s",
-                    string);
-
-                if (temp[0] != '/')
-                    snprintf(temp, sizeof(temp),
-                        "%s/%s",
-                        session->path_target,
-                        string);
-
-                length = session->ignores++;
-                session->ignore[length] = strdup(temp);
+                add_ignore(session,string);   
             }
 
             ucl_object_iterate_free(iti);
@@ -3541,6 +3575,7 @@
         .remote_data_file    = NULL,
         .remote_history_file = NULL,
         .ignore              = NULL,
+        .reg_ignore         = NULL,
         .ignores             = 0,
         .commit_history      = false,
         .keep_pack_file      = false,
@@ -3581,7 +3616,7 @@
 
     /* Process the command line parameters. */
 
-    while ((option = getopt(argc, argv, "C:cd:h:klrS:t:u:v:w:")) != -1) {
+    while ((option = getopt(argc, argv, "I:C:cd:h:klrS:t:u:v:w:")) != -1) {
         switch (option) {
             case 'C':
                 if (session.verbosity)
@@ -3589,6 +3624,9 @@
                         "# Configuration file: %s\n",
                         configuration_file);
                 break;
+            case 'I':
+                add_ignore(&session,optarg);
+                break;   
             case 'c':
                 session.clone = true;
                 break;
@@ -4011,9 +4049,10 @@
             "important changes.\n%s#\n",
             session.updating);
 
-    for (x = 0; x < session.ignores; x++)
+    for (x = 0; x < session.ignores; x++) {
         free(session.ignore[x]);
-
+                free(session.reg_ignore[x]);
+                 }
     free(session.ignore);
     free(session.response);
     free(session.object);
 
OP
J

jmehr

Member

Reaction score: 59
Messages: 32

here is a patch that adds regex support
also adds -I ignore on command line (works like ignore in config file)
a string preceded with a percent sign (%) is considered a regex otherwise a path
i had to modify prune_tree to check for ignores, otherwise you had to ignore work work/port work/port/file to protect a modified file
the only test i've done is
./gitup -v1 -I "%gitup/work/.*/gitup.c$" ports

Diff:
--- work/gitup-0.96/gitup.c    2021-09-05 19:58:01.000000000 +0300
+++ /home/titus/builds/gitup.c    2021-12-01 12:38:45.404988000 +0200
@@ -52,6 +52,7 @@
 #include <string.h>
 #include <unistd.h>
 #include <zlib.h>
+#include <regex.h>
 
 #define GITUP_VERSION     "0.96"
 #define BUFFER_UNIT_SMALL  4096
@@ -122,6 +123,7 @@
     char                *remote_data_file;
     char                *remote_history_file;
     char               **ignore;
+    regex_t           **reg_ignore;              
     int                  ignores;
     bool                 keep_pack_file;
     bool                 use_pack_file;
@@ -381,11 +383,18 @@
             return (false);
 
     for (x = 0; x < session->ignores; x++) {
+        if(!session->reg_ignore[x]) {
           ignore = session->ignore[x];
             
          if (strncmp(path, ignore, strlen(ignore)) == 0)
                return (true);
+         } else {
+           if(!regexec(session->reg_ignore[x],path,0,NULL,0)) {
+               printf("skipped %s with %s\n",path,session->ignore[x]);
+               return (true);
                }
+          }  
+    }
 
     return (false);
 }
@@ -511,6 +520,8 @@
 
             prune_tree(session, full_path);
         } else {
+    
+          if(!ignore_file(session,full_path,IGNORE_DELETE))
             if (remove(full_path) == -1)
                 err(EXIT_FAILURE,
                     "prune_tree: cannot remove %s",
@@ -3073,6 +3084,50 @@
 }
 
 
+static void
+add_ignore(connector *session,const char *string)
+{
+    char                temp[BUFFER_UNIT_SMALL];
+    int    ret_temp, length;
+    regex_t reg_temp;
+
+    session->ignore = (char **)realloc(
+        session->ignore,
+        (session->ignores + 1)*sizeof(char *));
+
+    session->reg_ignore = (regex_t **)realloc(
+        session->reg_ignore,
+        (session->ignores + 1)*sizeof(regex_t *));
+    if (session->ignore == NULL || session->reg_ignore == NULL)
+        err(EXIT_FAILURE,
+            "load_config_section: malloc");
+    session->reg_ignore[session->ignores] = NULL;    
+    snprintf(temp, sizeof(temp),
+        "%s",
+        string);
+
+    if (temp[0] != '/')
+        snprintf(temp, sizeof(temp),
+            "%s/%s",
+            session->path_target,
+            string);
+    if(string[0] == '%') {
+      ret_temp = regcomp(&reg_temp,string + 1,REG_EXTENDED);
+      if(ret_temp) {
+       warnx("warning: can't compile %s, ignoring\n",string + 1);
+       } else {
+       session->reg_ignore[session->ignores] = malloc(sizeof(regex_t));
+       if(!session->reg_ignore[session->ignores]) {
+       err(EXIT_FAILURE,
+            "load_config_section: malloc2");
+        }
+       memcpy(session->reg_ignore[session->ignores],&reg_temp,sizeof(regex_t));
+       }
+     }
+    length = session->ignores++;
+    session->ignore[length] = strdup(temp);
+    
+}
 /*
  * load_config_section
  *
@@ -3088,7 +3143,6 @@
     char                temp[BUFFER_UNIT_SMALL];
     int                 integer, length = 0;
     bool                boolean, target, path, ignores;
-
     its = ucl_object_iterate_new(section);
 
     while ((pair = ucl_object_iterate_safe(its, true))) {
@@ -3154,27 +3208,7 @@
 
             while ((ignore = ucl_object_iterate_safe(iti, true))) {
                 string = ucl_object_tostring(ignore);
-
-                session->ignore = (char **)realloc(
-                    session->ignore,
-                    (session->ignores + 1)*sizeof(char *));
-
-                if (session->ignore == NULL)
-                    err(EXIT_FAILURE,
-                        "load_config_section: malloc");
-
-                snprintf(temp, sizeof(temp),
-                    "%s",
-                    string);
-
-                if (temp[0] != '/')
-                    snprintf(temp, sizeof(temp),
-                        "%s/%s",
-                        session->path_target,
-                        string);
-
-                length = session->ignores++;
-                session->ignore[length] = strdup(temp);
+                add_ignore(session,string);
             }
 
             ucl_object_iterate_free(iti);
@@ -3541,6 +3575,7 @@
         .remote_data_file    = NULL,
         .remote_history_file = NULL,
         .ignore              = NULL,
+        .reg_ignore         = NULL,
         .ignores             = 0,
         .commit_history      = false,
         .keep_pack_file      = false,
@@ -3581,7 +3616,7 @@
 
     /* Process the command line parameters. */
 
-    while ((option = getopt(argc, argv, "C:cd:h:klrS:t:u:v:w:")) != -1) {
+    while ((option = getopt(argc, argv, "I:C:cd:h:klrS:t:u:v:w:")) != -1) {
         switch (option) {
             case 'C':
                 if (session.verbosity)
@@ -3589,6 +3624,9 @@
                         "# Configuration file: %s\n",
                         configuration_file);
                 break;
+            case 'I':
+                add_ignore(&session,optarg);
+                break;
             case 'c':
                 session.clone = true;
                 break;
@@ -4011,9 +4049,10 @@
             "important changes.\n%s#\n",
             session.updating);
 
-    for (x = 0; x < session.ignores; x++)
+    for (x = 0; x < session.ignores; x++) {
         free(session.ignore[x]);
-
+                free(session.reg_ignore[x]);
+                 }
     free(session.ignore);
     free(session.response);
     free(session.object);

Thank you! I committed a simplified version yesterday (everything in the ignores list goes through regcomp() and if we run into any problems, we can add a tuneable to include the REG_NOSPEC flag).

I also added a commented-out entry to the ports ignores list in gitup.conf:

"/work/|/work$"

which will preserve any /work/ directories in your ports tree.
 

aafbsd

New Member

Reaction score: 4
Messages: 15

When using an ignore entry which matches a self created file (e.g. "sysutils/bareos17-server/lala")
and a subsequent gitup removes "sysutils/bareos17-server", our to-be-ignored file is removed
as well. This can be circumvented with the following patch (warning: copy and paste):

--- gitup.c.ORI 2021-12-07 09:47:08.645516000 +0100 +++ gitup.c 2021-12-07 09:47:02.493263000 +0100 @@ -514,6 +514,8 @@ "prune_tree: cannot stat() %s", full_path); + if( ignore_file( session, full_path, IGNORE_DELETE )) + continue; if (S_ISDIR(sb.st_mode) != 0) { if ((entry->d_namlen == 1) && (strcmp(entry->d_name, "." ) == 0)) continue;

Of course this results in a warning

" ! cannot remove /tmp/g/sysutils/bareos17-server"

and has to be dealt with manually but it's still better than silently deleting it.
 
OP
J

jmehr

Member

Reaction score: 59
Messages: 32

When using an ignore entry which matches a self created file (e.g. "sysutils/bareos17-server/lala")
and a subsequent gitup removes "sysutils/bareos17-server", our to-be-ignored file is removed
as well. This can be circumvented with the following patch (warning: copy and paste):

--- gitup.c.ORI 2021-12-07 09:47:08.645516000 +0100 +++ gitup.c 2021-12-07 09:47:02.493263000 +0100 @@ -514,6 +514,8 @@ "prune_tree: cannot stat() %s", full_path); + if( ignore_file( session, full_path, IGNORE_DELETE )) + continue; if (S_ISDIR(sb.st_mode) != 0) { if ((entry->d_namlen == 1) && (strcmp(entry->d_name, "." ) == 0)) continue;

Of course this results in a warning

" ! cannot remove /tmp/g/sysutils/bareos17-server"

and has to be dealt with manually but it's still better than silently deleting it.

I'm not having any luck recreating the scenario you've described. When I create custom directories and files and add them to the ignores list, they survive subsequent runs.

When I clone an old commit of the https://github.com/freebsd/freebsd-ci repository where scripts/build/config-11 still exists, add a custom directory scripts/build/config-11/stuff to it and add another file scripts/build/config-11/stuff/test, add "scripts/build/config-11/stuff" to the ignores list and then pull the latest commit, the contents of scripts/build/config-11 are removed except for the custom directory and file.

Could you please send me a specific case where the custom files are deleted? Thanks!
 

covacat

Daemon

Reaction score: 801
Messages: 1,465

if you ignore a file in a non repository dir eg /usr/ports/archivers/zip/work/file.c it will get nuked because /usr/ports/archivers/zip/work/
will be checked first and pruned
 

aafbsd

New Member

Reaction score: 4
Messages: 15

I'm not having any luck recreating the scenario you've described. When I create custom directories and files and add them to the ignores list, they survive subsequent runs.

When I clone an old commit of the https://github.com/freebsd/freebsd-ci repository where scripts/build/config-11 still exists, add a custom directory scripts/build/config-11/stuff to it and add another file scripts/build/config-11/stuff/test, add "scripts/build/config-11/stuff" to the ignores list and then pull the latest commit, the contents of scripts/build/config-11 are removed except for the custom directory and file.

Could you please send me a specific case where the custom files are deleted? Thanks!
I have an old (as of 2021/12/03) ports directory in /tmp/g. In my gitup.conf I have an
"ignores" entry including "math/eigen2/bla". And I have this file:

buildbox:/tmp>ll g/math/eigen2/bla
-rw-r----- 1 user wheel 2852 15 Dec 06:40 g/math/eigen2/bla

Now I run gitup (where the eigen2 port is gone) and my file gets deleted:

...
- /tmp/g/lang/ruby27/files/patch-include_ruby_ruby.h
- /tmp/g/math/eigen2
- /tmp/g/math/eigen2/Makefile
- /tmp/g/math/eigen2/distinfo
- /tmp/g/math/eigen2/pkg-descr
- /tmp/g/math/eigen2/pkg-plist
- /tmp/g/misc/freebsd-release-manifests/files/MANIFESTS/amd64-amd64-12.3-RC1
...

buildbox:/tmp>ll g/math/eigen2/bla
ls: g/math/eigen2/bla: No such file or directory

With my patch from above the file won't get touched (and the dir not deleted). This is with gitup
from ports, but including the fix from https://forums.freebsd.org/posts/542233.

So in order to reproduce:

1. checkout ports with gitup
2. tar it to some place
3. run gitup until you find some port getting deleted
4. restore from the tar
5. copy something to "category/deletedport/myfile"
6. add "category/deletedport/myfile" to "ignores"
7. run gitup
 
OP
J

jmehr

Member

Reaction score: 59
Messages: 32

I have an old (as of 2021/12/03) ports directory in /tmp/g. In my gitup.conf I have an
"ignores" entry including "math/eigen2/bla". And I have this file:

buildbox:/tmp>ll g/math/eigen2/bla
-rw-r----- 1 user wheel 2852 15 Dec 06:40 g/math/eigen2/bla

Now I run gitup (where the eigen2 port is gone) and my file gets deleted:

...
- /tmp/g/lang/ruby27/files/patch-include_ruby_ruby.h
- /tmp/g/math/eigen2
- /tmp/g/math/eigen2/Makefile
- /tmp/g/math/eigen2/distinfo
- /tmp/g/math/eigen2/pkg-descr
- /tmp/g/math/eigen2/pkg-plist
- /tmp/g/misc/freebsd-release-manifests/files/MANIFESTS/amd64-amd64-12.3-RC1
...

buildbox:/tmp>ll g/math/eigen2/bla
ls: g/math/eigen2/bla: No such file or directory

With my patch from above the file won't get touched (and the dir not deleted). This is with gitup
from ports, but including the fix from https://forums.freebsd.org/posts/542233.

So in order to reproduce:

1. checkout ports with gitup
2. tar it to some place
3. run gitup until you find some port getting deleted
4. restore from the tar
5. copy something to "category/deletedport/myfile"
6. add "category/deletedport/myfile" to "ignores"
7. run gitup
I'm still not having any luck reproducing the issue. I have a copy of the ports tree from commit b298802b56c2f85df0cc28426b8612ea01751bd8 on December 5 (which has the math/eigen2 port) and if I add "math/eigen2/bla" to the ignores list, execute touch math/eigen2/bla and then run gitup ports to fetch the latest revision, the contents of math/eigen2 are removed except for math/eigen2/bla.

Could you please let me see the contents of your gitup.conf and could you try using the latest revision from gitup's repository? Thanks!
 

covacat

Daemon

Reaction score: 801
Messages: 1,465

add math/eigen2/bar/bla.txt and exclude the file bla.txt
it will be erased because math/eigen2/bar comes first, it is not excluded and it is pruned (which deletes bla.txt)
 
OP
J

jmehr

Member

Reaction score: 59
Messages: 32

add math/eigen2/bar/bla.txt and exclude the file bla.txt it will be erased because math/eigen2/bar comes first, it is not excluded and it is pruned (which deletes bla.txt)

I still can't reproduce it but I'm thinking this might be due to differences in the ordering of files returned by the opendir() and readdir() calls. I just committed aafbsd's patch (thank you!). How does it work for both of you now?
 

grahamperrin

Beastie's Twin

Reaction score: 1,716
Messages: 5,101

jmehr please: is option -c intended for cases where, for example, a repair fails?

Spun off from <https://forums.freebsd.org/posts/549178> and with reference to the attached file, I had a peculiar experience. It seemed that:
  • gitup -c release (15:04, 549 in history) followed, not immediately, by gitup release (15:20, 560) resulted in repair, and a prompt to re-run.
It's possible that I keyed Control-C during the first of the commands, although I can't imagine doing so. The linked post confirms that the first command ran to completion; Done.

Can you think of any other explanation for the second command resulting in a repair?



Below, I can not reproduce the peculiarity.

Code:
root@mowa219-gjp4-vm-freebsd-13-zfs:~ # gitup -V
gitup version 0.96
root@mowa219-gjp4-vm-freebsd-13-zfs:~ # freebsd-version -kru
13.0-RELEASE-p4
13.0-RELEASE-p4
13.0-RELEASE-p5
root@mowa219-gjp4-vm-freebsd-13-zfs:~ # pkg -vv | grep -e url -e enabled
    url             : "pkg+http://pkg.FreeBSD.org/FreeBSD:13:amd64/latest",
    enabled         : yes,
    url             : "https://alpha.pkgbase.live/current/FreeBSD:13:amd64/latest",
    enabled         : no,
    url             : "file:///usr/local/poudriere/data/packages/13-default",
    enabled         : yes,
root@mowa219-gjp4-vm-freebsd-13-zfs:~ # gitup -c release
# Scanning local repository...
# Host: git.freebsd.org
# Port: 443
# Repository Path: /src.git
# Target Directory: /usr/src
# Have: 2646dd665909e60a369015c17cb602515e6025dc
# Want: 2646dd665909e60a369015c17cb602515e6025dc
# Branch: releng/13.0
# Done.
root@mowa219-gjp4-vm-freebsd-13-zfs:~ # gitup release
# Scanning local repository...
# Host: git.freebsd.org
# Port: 443
# Repository Path: /src.git
# Target Directory: /usr/src
# Have: 2646dd665909e60a369015c17cb602515e6025dc
# Want: 2646dd665909e60a369015c17cb602515e6025dc
# Branch: releng/13.0
# Done.
root@mowa219-gjp4-vm-freebsd-13-zfs:~ # grep -v \# /etc/freebsd-update.conf | sort


















Components src world kernel
IDSIgnorePaths /usr/share/man/cat
IDSIgnorePaths /usr/share/man/whatis
IDSIgnorePaths /var/db/locate.database
IDSIgnorePaths /var/log
IgnorePaths
KeyPrint 800651ef4b4c71c27e60786d7b487188970f4b4169cc055784e21eb71d410cc5
MergeChanges /etc/ /boot/device.hints
ServerName update.FreeBSD.org
UpdateIfUnmodified /etc/ /var/ /root/ /.cshrc /.profile
root@mowa219-gjp4-vm-freebsd-13-zfs:~ # setenv PAGER cat
root@mowa219-gjp4-vm-freebsd-13-zfs:~ # freebsd-update fetch install
Looking up update.FreeBSD.org mirrors... 2 mirrors found.
Fetching metadata signature for 13.0-RELEASE from update2.freebsd.org... done.
Fetching metadata index... done.
Inspecting system... done.
Preparing to download files... done.

No updates needed to update system to 13.0-RELEASE-p5.
No updates are available to install.
root@mowa219-gjp4-vm-freebsd-13-zfs:~ # history -S
root@mowa219-gjp4-vm-freebsd-13-zfs:~ # history | tail -n 21
   556  15:16   etcupdate -B
…
root@mowa219-gjp4-vm-freebsd-13-zfs:~ # history | tail -n 29
   549  15:04   gitup -c release
   550  15:04   setenv PAGER cat ; freebsd-update fetch install
   551  15:10   cat /etc/freebsd-update.conf
   552  15:11   ls -hl /etc/freebsd-update.conf
   553  15:11   grep -v \# /etc/freebsd-update.conf
   554  15:11   ls -hl /etc/freebsd-update.conf
   555  15:11   grep -v \# /etc/freebsd-update.conf | sort
   556  15:16   etcupdate -B
   557  15:16   etcupdate -B
   558  15:16   etcupdate -p
   559  15:16   etcupdate resolve
   560  15:20   gitup release
   561  15:24   history | grep gitup
   562  15:24   history -S
   563  15:30   pkg install qterminal
   564  15:33   gitup -v
   565  15:33   gitup --version
   566  15:33   man gitup
   567  15:34   gitup -V
   568  15:34   freebsd-version -kru
   569  15:34   pkg -vv | grep -e url -e enabled
   570  15:34   gitup -c release
   571  15:41   gitup release
   572  15:42   grep -v \# /etc/freebsd-update.conf | sort
   573  15:43   setenv PAGER cat
   574  15:43   freebsd-update fetch install
   575  15:51   history -S
   576  15:53   history | tail -n 21
   577  15:54   history | tail -n 29
root@mowa219-gjp4-vm-freebsd-13-zfs:~ #

<https://cgit.freebsd.org/src/log/?h=releng/13.0> as expected, nothing since 2021-11-03.

If you wonder about the four unnecessary etcupdate(8) commands:
  • I was clutching at straws (in the context of the other topic, freebsd-update fetching but failing to install updates)
  • they seem to have no impact.



The screenshot below is to remind me that I have a VirtualBox snapshot of the machine as it was thirty days ago.

jmehr I'm inclined to treat the peculiarity above as negligible, a one-off without explanation. Alternatively: if you feel that it deserves further investigation, I might find time later in the month to thrash around with the 2021-12-03 snapshot as a basis.

1641141425012.png
 

Attachments

  • gitup partial history.txt
    40.1 KB · Views: 26

grahamperrin

Beastie's Twin

Reaction score: 1,716
Messages: 5,101

It's possible that I keyed Control-C during the first of the commands, although I can't imagine doing so. The linked post confirms that the first command ran to completion; Done.

More specifically, quoting from the linked post:

Code:
root@mowa219-gjp4-vm-freebsd-13-zfs:~ # gitup -c release                                     # Scanning local repository...
# Host: git.freebsd.org
# Port: 443
# Repository Path: /src.git
# Target Directory: /usr/src
# Have: 2646dd665909e60a369015c17cb602515e6025dc
# Want: 2646dd665909e60a369015c17cb602515e6025dc
# Branch: releng/13.0
# Done.
root@mowa219-gjp4-vm-freebsd-13-zfs:~ # …
  • after the word Done appeared, there was a long wait
  • I did wonder whether something had got stuck
  • I might have impatiently keyed Control-C to regain the command prompt.
Please: what exactly happens during that waiting period after the word Done appears?
 

astyle

Daemon

Reaction score: 1,021
Messages: 2,017

A possibility that comes to my mind is that gitup(1) could be not exiting very gracefully, or the machine was waiting on a different mutex before deciding that sh and gitup merit attention from the processor. As in, there could be a hung process somewhere. Try running # ps -ax in a different terminal/tty to verify my assumptions.
 

aafbsd

New Member

Reaction score: 4
Messages: 15

I still can't reproduce it but I'm thinking this might be due to differences in the ordering of files returned by the opendir() and readdir() calls. I just committed aafbsd's patch (thank you!). How does it work for both of you now?
I don't have that old test tree anymore but I just created a new one and will see what happens
in the next days.

BTW, I had trouble regarding the new regex feature: The man pages simply say:

"Regular expressions are supported"

So, is everything here treated as regex or do they have to be prefixed (and optionally postfixed) with
something (e.g. '/')? This should be clarified, especially for people coming from older versions...
In my case, I didn't know if I have to escape the '.' or not :).
 
Top