close a breach that could lead to optimistic done

Manifest

@@ -164,6 +164,7 @@ unit_tests/critical2.py
 unit_tests/cwd.py
 unit_tests/cycle.py
 unit_tests/depend.py
+unit_tests/depend2.py
 unit_tests/depend_verbose.py
 unit_tests/dup.py
 unit_tests/dyn_accesses.py
@@ -177,6 +178,7 @@ unit_tests/hard_lnks.py
 unit_tests/hello_world.py
 unit_tests/hide.py
 unit_tests/home.py
+unit_tests/hot.py
 unit_tests/ignore.py
 unit_tests/ignore_err.py
 unit_tests/jemalloc.py

TO_DO

@@ -13,7 +13,6 @@ items :
 ****************************************************************************************************
 
 * crash if python() dynamic attribute returns None
-* crash in regexpr (cf mail NM 2024/04/08 21:29)
 * fix compilation with LMAKE_FLAGS=ST
 * missing some deps when reading elf
 	- it seems that libc.so is missing at least in some occasions

_bin/sys_config

@@ -61,7 +61,7 @@ CXX=${CXX:-g++}
 type -p $CXX >/dev/null    || { echo cannot find      c++ compiler $CXX ; exit 1 ; }
 
 case "$($CXX --version|head -1)" in
-	*clang* ) CXX_FLAVOR=clang ; v=$($CXX -dumpversion) ; [ ${v%%.*} -lt 15 ] && { echo clang version must be at least 15  ; exit 1 ; } ;;
+	*clang* ) CXX_FLAVOR=clang ; v=$($CXX -dumpversion) ; [ ${v%%.*} -lt 14 ] && { echo clang version must be at least 14  ; exit 1 ; } ;;
 	*g++*   ) CXX_FLAVOR=gcc   ; v=$($CXX -dumpversion) ; [ ${v%%.*} -lt 11 ] && { echo gcc   version must be at least 11  ; exit 1 ; } ;;
 	*       )                                                                      echo cannot recognize c++ compiler $CXX ; exit 1     ;;
 esac

_lib/lmake/__init__.py

@@ -45,54 +45,55 @@ def version(major,minor) :
 #_group     = _os.getgroups()[0]
 
 config = pdict(
-	heartbeat        = 10             # in seconds, minimum interval between 2 heartbeat checks (and before first one) for the same job (no heartbeat if None)
-,	heartbeat_tick   =  0.1           # in seconds, minimum internval between 2 heartbeat checks (globally)                             (no heartbeat if None)
-,	link_support     = 'Full'         # symlinks are supported. Other values are 'None' (no symlink support) or 'File' (symlink to file only support)
-#,	local_admin_dir  = 'LMAKE_LOCAL'  # directory in which to store data that are private to the server (not accessed by remote executing hosts) (default is within LMAKE dir)
-,	max_dep_depth    = 1000           # used to detect infinite recursions and loops
-,	max_error_lines  = 100            # used to limit the number of error lines when not reasonably limited otherwise
-,	network_delay    = 1              # delay between job completed and server aware of it. Too low, there may be spurious lost jobs. Too high, tool reactivity may rarely suffer.
-,	path_max         = 400            # max path length, but a smaller value makes debugging easier (by default, not activated)
-#,	reliable_dirs    = False          # if true, close to open coherence is deemed to encompass enclosing directory coherence (improve performances)
-#	                                  # - forced true if only local backend is used
-#	                                  # - set   true  for ceph
-#	                                  # - leave false for NFS
-#,	remote_admin_dir = 'LMAKE_REMOTE' # directory in which to store job trace during remote job execution                                      (default is within LMAKE dir)
-#,	remote_tmp_dir   = 'LMAKE_TMP'    # directory in which to store tmp data  during remote job execution (not used when keep_tmp is enforced) (default is within LMAKE dir)
-#,	rules_module     = 'rules'        # module to import to define rules  . By default, rules are directly defined in Lmakefile.py
-#,	sources_module   = 'sources'      # module to import to define sources. By default, 'lmake.auto_sources' which lists files in Manifest or searches git (recursively) if lmake.sources is not set
-,	sub_prio_boost   = 1              # increment to add to rules defined in sub-repository (multiplied by directory depth of sub-repository) to boost local rules
-,	console = pdict(                  # tailor output lines
-		date_precision = None         # number of second decimals in the timestamp field
-	,	host_length    = None         # length of the host field (lines will be misaligned if a host is longer)
-	,	has_exec_time  = True         # if True, output the exec_time field
+	disk_date_precision = 0.010          # in seconds, precisions of dates on disk, must account for date granularity and date discrepancy between executing hosts and disk servers
+,	heartbeat           = 10             # in seconds, minimum interval between 2 heartbeat checks (and before first one) for the same job (no heartbeat if None)
+,	heartbeat_tick      =  0.1           # in seconds, minimum internval between 2 heartbeat checks (globally)                             (no heartbeat if None)
+,	link_support        = 'Full'         # symlinks are supported. Other values are 'None' (no symlink support) or 'File' (symlink to file only support)
+#,	local_admin_dir     = 'LMAKE_LOCAL'  # directory in which to store data that are private to the server (not accessed by remote executing hosts) (default is within LMAKE dir)
+,	max_dep_depth       = 1000           # used to detect infinite recursions and loops
+,	max_error_lines     = 100            # used to limit the number of error lines when not reasonably limited otherwise
+,	network_delay       = 1              # delay between job completed and server aware of it. Too low, there may be spurious lost jobs. Too high, tool reactivity may rarely suffer.
+,	path_max            = 400            # max path length, but a smaller value makes debugging easier (by default, not activated)
+#,	reliable_dirs       = False          # if true, close to open coherence is deemed to encompass enclosing directory coherence (improve performances)
+#	                                     # - forced true if only local backend is used
+#	                                     # - set   true  for ceph
+#	                                     # - leave false for NFS
+#,	remote_admin_dir    = 'LMAKE_REMOTE' # directory in which to store job trace during remote job execution                                      (default is within LMAKE dir)
+#,	remote_tmp_dir      = 'LMAKE_TMP'    # directory in which to store tmp data  during remote job execution (not used when keep_tmp is enforced) (default is within LMAKE dir)
+#,	rules_module        = 'rules'        # module to import to define rules  . By default, rules are directly defined in Lmakefile.py
+#,	sources_module      = 'sources'      # module to import to define sources. By default, 'lmake.auto_sources' which lists files in Manifest or searches git (recursively) if lmake.sources is not set
+,	sub_prio_boost      = 1              # increment to add to rules defined in sub-repository (multiplied by directory depth of sub-repository) to boost local rules
+,	console = pdict(                     # tailor output lines
+		date_precision = None            # number of second decimals in the timestamp field
+	,	host_length    = None            # length of the host field (lines will be misaligned if a host is longer)
+	,	has_exec_time  = True            # if True, output the exec_time field
 	)
-,	n_tokens_tab = pdict()            # table of number of tokens referenced by rules. This indirection allows dynamic update of this value while rules cannot be dynamically updated
-,	backends = pdict(                 # PER_BACKEND : provide a default configuration for each backend
-		precisions = pdict(           # precision of resources allocated for jobs, one entry for each standard resource (for all backends).
-			cpu = 4                   # encodes the highest number with full granularity, 4 is a reasonable value
-		,	mem = 4                   # 4 means possible values are 1 2, 3, 4, 6, 8, 10, 12, 14, 16, 20, 24, 28, 32, 40...
-		,	tmp = 4                   # 8 would mean possible values are 1 2, 3, 4, 5, 6, 7, 8, 10, ...
+,	n_tokens_tab = pdict()               # table of number of tokens referenced by rules. This indirection allows dynamic update of this value while rules cannot be dynamically updated
+,	backends = pdict(                    # PER_BACKEND : provide a default configuration for each backend
+		precisions = pdict(              # precision of resources allocated for jobs, one entry for each standard resource (for all backends).
+			cpu = 4                      # encodes the highest number with full granularity, 4 is a reasonable value
+		,	mem = 4                      # 4 means possible values are 1 2, 3, 4, 6, 8, 10, 12, 14, 16, 20, 24, 28, 32, 40...
+		,	tmp = 4                      # 8 would mean possible values are 1 2, 3, 4, 5, 6, 7, 8, 10, ...
 		)
-	,	local = pdict(                # entries mention the total availability of resources
-		#	interface = _interface    # address at which lmake can be contacted from jobs launched by this backend, can be :
-		#	                          # - ''                     : loop-back address (127.0.0.1) for local backend, hostname for remote backends
-		#	                          # - standard dot notation  : for example '192.168.0.1'
-		#	                          # - network interface name : the address of the host on this interface (as shown by ifconfig)
-		#	                          # - a host name            : the address of the host as found in networkd database (as shown by ping)
-		#	                          # - default is loopback for local backend and hostname for the others
-			cpu = _cpu                # total number of cpus available for the process, and hence for all jobs launched locally
-		,	mem = str(_mem>>20)+'M'   # total available memory in MBytes
-		,	tmp = 0                   # total available temporary disk space in MBytes
+	,	local = pdict(                   # entries mention the total availability of resources
+		#	interface = _interface       # address at which lmake can be contacted from jobs launched by this backend, can be :
+		#	                             # - ''                     : loop-back address (127.0.0.1) for local backend, hostname for remote backends
+		#	                             # - standard dot notation  : for example '192.168.0.1'
+		#	                             # - network interface name : the address of the host on this interface (as shown by ifconfig)
+		#	                             # - a host name            : the address of the host as found in networkd database (as shown by ping)
+		#	                             # - default is loopback for local backend and hostname for the others
+			cpu = _cpu                   # total number of cpus available for the process, and hence for all jobs launched locally
+		,	mem = str(_mem>>20)+'M'      # total available memory in MBytes
+		,	tmp = 0                      # total available temporary disk space in MBytes
 		)
 	)
-,	caches = pdict(                   # PER_CACHE : provide an explanation for each cache method
-	#	dir = pdict(                  # when rule specifies cache = 'dir' , this cache is selected
-	#		tag    = 'dir'            # specify the caching method, must be one of the supported method
-	#	,	repo   = root_dir         # an id that identifies the repository, no more than one entry is stored in the cache for a given job and tag
-	#	,	dir    = '/cache_dir'     # the directory in which cached results are stored
-	#	,	size   = 10<<30           # the overall size of this cache
-	#	,	group  = _group           # the group used to write to the cache. If user does not belong to this group, read-only access is still possible
+,	caches = pdict(                      # PER_CACHE : provide an explanation for each cache method
+	#	dir = pdict(                     # when rule specifies cache = 'dir' , this cache is selected
+	#		tag    = 'dir'               # specify the caching method, must be one of the supported method
+	#	,	repo   = root_dir            # an id that identifies the repository, no more than one entry is stored in the cache for a given job and tag
+	#	,	dir    = '/cache_dir'        # the directory in which cached results are stored
+	#	,	size   = 10<<30              # the overall size of this cache
+	#	,	group  = _group              # the group used to write to the cache. If user does not belong to this group, read-only access is still possible
 	#	)
 	)
 ,	colors = pdict(

doc/lmake.texi

@@ -1281,6 +1281,15 @@ Recognized attributes are :
 @multitable @columnfractions 0.1 0.07 0.05 0.78
 @headitem Attribute @tab Default @tab Update @tab Description
 
+@item @code{disk_date_precision}
+@tab @code{0.010}
+@tab Static
+@tab This attribute instruct @lmake to take some margin (expressed in seconds) when it must rely on file dates to decide about event orders.
+It must account for file date granularity (generally a few ms) and date discrepancy between executing hosts and disk servers (generally a few ms when using NTP).
+If too low, there are risks that @lmake consider that data read by a job are up to date while they have been modified shortly after.
+If too high, there is a small impact on performance as @lmake will consider out of date data that are actually up to date.
+The default value should be safe in usual cases and user should hardly need to modify it.
+
 @item @code{heartbeat}
 @tab @code{10}
 @tab Static

docker/ubuntu22.docker

@@ -15,6 +15,7 @@ RUN apt-get update --fix-missing       # 1 : in case of failure, try to incremen
 RUN apt-get install -y build-essential
 RUN apt-get install -y ed g++-12 git texinfo libseccomp-dev
 RUN apt-get install -y python2 python3-dev python3-numba python3-venv
+RUN apt-get install -y clang
 
 RUN     useradd --uid 1001 --home-dir /home/cdy  cdy
 USER    1001:1001

src/autodep/clmake.cc

@@ -85,7 +85,7 @@ static PyObject* depend( PyObject* /*null*/ , PyObject* args , PyObject* kwds )
 		Ptr<Dict> res { New } ;
 		if (+files) {           // fast path : else, depend on no files
 			//
-			JobExecRpcReply reply = _g_record.direct(JobExecRpcReq( Proc::DepInfos , ::copy(files) , ad , no_follow , true/*sync*/ , "depend" )) ;
+			JobExecRpcReply reply = _g_record.direct(JobExecRpcReq( Proc::DepVerbose , ::copy(files) , ad , no_follow , true/*sync*/ , "depend" )) ;
 			//
 			SWEAR( reply.dep_infos.size()==files.size() , reply.dep_infos.size() , files.size() ) ;
 			for( size_t i=0 ; i<reply.dep_infos.size() ; i++ ) {