Re: [PATCH] Implementing IncludeCmd (feature request #17)
On Thu, Aug 31, 2023 at 10:58:27AM +0200, Michiel van den Heuvel wrote:
so if you think it would be a good use of *your* time, then go ahead.
After some somewhat more synchronous communication, it might be.
are you asking for a chat channel?
i'm @ossi:kde.org on matrix, ossi on libera.chat irc (if the matrix
bridge feels like working), and @ossilator on various social media and
chat sites.
i could create an official project channel (on matrix?), but
historically it seems a bit ... superfluous.
On the subject of more things I could do, is there a way to reproduce
the 2 regressions?
congrats, you forced me to actually research what the regressions are.
*grumble* :-D
for the imap thing, i suspect any mail with a large enough attachment
will do. maybe the imap compress extension needs to be in use, and/or
tls. no idea.
(possibly) related threads:
https://sourceforge.net/p/isync/mailman/isync-devel/thread/Y00lhIm7VdrJzG/D%40ugly/#msg37721793
https://sourceforge.net/p/isync/mailman/isync-devel/thread/87h740x2xe.fsf%40wavexx.thregr.org/#msg37675548
https://sourceforge.net/p/isync/mailman/isync-devel/thread/87edk45p9o.fsf%40b3l.xyz/#msg37883904
for the other problem, there is
https://sourceforge.net/p/isync/mailman/isync-devel/thread/f4d61b60-7a93-6b78-90d1-b96b285caa9c%40quaternum.net/#msg37708463
which contains a speculative analysis.
there is https://sourceforge.net/p/isync/bugs/69/ which may be actually
the same thing (sounds similar, but i'm not going to think about it at
this time of day). it has a patch attached, which is certainly
under-tested, and needs a sanity check, possibly resulting in a complete
rewrite.
speaking of tested, if you _really_ wanted to go overboard, you could
add a proper unit testing suite. so far only the core syncing algo is
tested, in a way that is really an integration test (which has the
advantage of testing a bit more as a side effect, but it's slow and
inelegant). this would also nicely fit with formalizing the error
reporting.
Subject: [PATCH] Add IncludeCmd directive to config parser
+static char *
+check_excess_tokens( conffile_t *cfile )
+{
+ char *arg = NULL;
+
i generally prefer early returns (and even goto) over temporaries.
int
getcline( conffile_t *cfile )
{
...
+ else if ((arg = check_excess_tokens( cfile )))
+ conf_error( cfile, "excess token '%s', not executing
"
+ "potentially malformed command\n",
arg );
my thinking was that the error message would be printed already by the
check function (hence "check", not "get"). i don't think the extra
verbosity about the specific consequence adds all that much value.
then the check function can just return an int.
oh, one thing i totally failed to notice: this is a noteworthy new
feature, so it should get an entry in NEWS. something like "Added
support for scripting arbitrary parts of the config file.", before the
currently last two entries, i think.
and you should add yourself to the end of the main contributor list in
AUTHORS.
regards
___
isync-devel mailing list
isync-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/isync-devel
Re: Restore after crash
My situation is actually the oppisiate. My Ubuntu machine was running 1.3 and the mint machine is running 1.4. But I did copy over the .mbsync directory. That is why I asked if I needed to redownload all my mail. --- Thanks, Robert Bower W9RWB WRPH745 On 2023-08-31 01:02, Oswald Buddenhagen wrote: On Wed, Aug 30, 2023 at 11:16:47AM -0700, Robert Bower wrote: Error: incompatible journal version (got 3, expected 4) it means you downgraded isync. you need to build a newer version (1.4) yourself if there are no backports for your mint version. regards ___ isync-devel mailing list isync-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/isync-devel ___ isync-devel mailing list isync-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/isync-devel
Re: Restore after crash
On Thu, Aug 31, 2023 at 06:54:06AM -0700, Robert Bower wrote: My situation is actually the oppisiate. My Ubuntu machine was running 1.3 and the mint machine is running 1.4. right, the journal handling is not backward compatible, unlike that of the proper state file. so you need to install the older version and run it (only!) over the mailbox(es) that didn't finish properly. afterwards everything should work fine with the new version. regards ___ isync-devel mailing list isync-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/isync-devel
Re: [PATCH] Implementing IncludeCmd (feature request #17)
> that depends on whose perspective you're assuming here.
> i'm likely not going to get it done anytime soon. pushing patches on me
> prompts me into action, so it's good for the project (not so much for my
> other projects, but the trade-off isn't that bad, as i'm actually much
> more efficient at reviewing than hacking myself).
> so if you think it would be a good use of *your* time, then go ahead.
After some somewhat more synchronous communication, it might be. This
project has been a nice refresher course on C.
> i implied that you need to do that anyway to handle things properly.
> also note that you'd save the duplicated check for the command.
>
> but looking at it again, duplication probably isn't even necessary: you
> can factor out check_excess_tokens() and make it also return a value.
I hope this is what you meant.
> the first patch is now Perfect (TM), so no need to re-send it.
Nice! Progress.
On the subject of more things I could do, is there a way to reproduce
the 2 regressions? I am not familiar with IMAP, but do know how to fix
a regression. I'm not in any hurry, and do not want to promise
anything. But I'd like to take a look.
Cheers, Michiel>From 8f4c551449e90f80cdbf31fd6a0b207e9802c3e7 Mon Sep 17 00:00:00 2001
From: Michiel van den Heuvel
Date: Thu, 17 Aug 2023 20:25:51 +0200
Subject: [PATCH] Add IncludeCmd directive to config parser
---
src/config.c | 88
src/config.h | 3 ++
src/mbsync.1 | 8 +
3 files changed, 93 insertions(+), 6 deletions(-)
diff --git a/src/config.c b/src/config.c
index ac90c25..a7476a9 100644
--- a/src/config.c
+++ b/src/config.c
@@ -61,12 +61,21 @@ expand_strdup( const char *s, const conffile_t *cfile )
}
}
+static void
+conf_print_loc( const conffile_t *cfile )
+{
+ if (cfile->eval_fp)
+ fprintf( stderr, "%s:%d:included:%d: ", cfile->file, cfile->line, cfile->eval_line );
+ else
+ fprintf( stderr, "%s:%d: ", cfile->file, cfile->line );
+}
+
void
conf_error( conffile_t *cfile, const char *fmt, ... )
{
va_list va;
- fprintf( stderr, "%s:%d: ", cfile->file, cfile->line );
+ conf_print_loc( cfile );
va_start( va, fmt );
vfprintf( stderr, fmt, va );
va_end( va );
@@ -79,7 +88,7 @@ conf_sys_error( conffile_t *cfile, const char *fmt, ... )
va_list va;
int errno_bak = errno;
- fprintf( stderr, "%s:%d: ", cfile->file, cfile->line );
+ conf_print_loc( cfile );
errno = errno_bak;
va_start( va, fmt );
vsys_error( fmt, va );
@@ -318,17 +327,72 @@ getopt_helper( conffile_t *cfile, int *cops, channel_conf_t *conf )
return 1;
}
+static void
+eval_cmd_popen( conffile_t *cfile, const char *cmd )
+{
+ if (!(cfile->eval_fp = popen( cmd, "r" ))) {
+ sys_error( "popen" );
+ cfile->err = 1;
+ return;
+ }
+ cfile->eval_line = 0;
+ cfile->eval_command = nfstrdup( cmd );
+}
+
+static void
+eval_cmd_pclose( conffile_t *cfile )
+{
+ int ret;
+
+ if ((ret = pclose( cfile->eval_fp ))) {
+ if (ret < 0) {
+ sys_error( "pclose" );
+ cfile->err = 1;
+ } else if (WIFSIGNALED( ret )) {
+ conf_error( cfile, "command \"%s\" crashed with signal %d\n",
+ cfile->eval_command, WTERMSIG( ret ) );
+ } else {
+ conf_error( cfile, "command \"%s\" exited with status %d\n",
+ cfile->eval_command, WEXITSTATUS( ret ) );
+ }
+ }
+ free( cfile->eval_command );
+ cfile->eval_fp = NULL;
+ cfile->eval_command = NULL;
+}
+
+static int
+read_cline( conffile_t *cfile )
+{
+ if (cfile->eval_fp) {
+ cfile->eval_line++;
+ if ((cfile->rest = fgets( cfile->buf, cfile->bufl, cfile->eval_fp )))
+ return 1;
+ eval_cmd_pclose( cfile );
+ }
+ cfile->line++;
+ return (cfile->rest = fgets( cfile->buf, cfile->bufl, cfile->fp )) != NULL;
+}
+
+static char *
+check_excess_tokens( conffile_t *cfile )
+{
+ char *arg = NULL;
+
+ if (cfile->rest)
+ arg = get_arg( cfile, ARG_OPTIONAL, NULL );
+ return arg;
+}
+
int
getcline( conffile_t *cfile )
{
char *arg;
int comment;
- if (cfile->rest && (arg = get_arg( cfile, ARG_OPTIONAL, NULL )))
+ if ((arg = check_excess_tokens( cfile )))
conf_error( cfile, "excess token '%s'\n", arg );
- while (fgets( cfile->buf, cfile->bufl, cfile->fp )) {
- cfile->line++;
- cfile->rest = cfile->buf;
+ while (read_cline( cfile )) {
if (!(cfile->cmd = get_arg( cfile, ARG_OPTIONAL, &comment ))) {
if (comment)
continue;
@@ -336,6 +400,16 @@ getcline( conffile_t *cfile )
}
if (!(cfile->val = get_arg( cfile, ARG_REQUIRED, NULL )))
continue;
+ if (!strcasecmp( cfile->cmd, "IncludeCmd" )) {
+ if (cfile->eval_fp)
+conf_error( cfile, "nested IncludeCmd\n" );
+ else if ((arg = check_excess_tokens( cfile )))
+conf_error( cfile, "excess token '%s', not executing "
+"potentially malformed command\n", arg );
+ else
+eval_cmd_popen( cfile, cfile->val );
+ continue;
+ }
return 1;
}
return 0;
@@ -488,6 +562,7 @@ load_config( const char *where )
return 1;
}
buf[sizeof(buf) - 1] = 0;
+
Re: Restore after crash
On Wed, Aug 30, 2023 at 11:16:47AM -0700, Robert Bower wrote: Error: incompatible journal version (got 3, expected 4) it means you downgraded isync. you need to build a newer version (1.4) yourself if there are no backports for your mint version. regards ___ isync-devel mailing list isync-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/isync-devel
