diff --git a/plugins/bm29/hplc.py b/plugins/bm29/hplc.py
index b5bf3ea..5e3a8da 100644
--- a/plugins/bm29/hplc.py
+++ b/plugins/bm29/hplc.py
@@ -10,7 +10,7 @@
 __contact__ = "Jerome.Kieffer@ESRF.eu"
 __license__ = "MIT"
 __copyright__ = "European Synchrotron Radiation Facility, Grenoble, France"
-__date__ = "21/02/2025" 
+__date__ = "27/05/2025"
 __status__ = "development"
 __version__ = "0.3.0"
 
@@ -51,7 +51,7 @@
 
 def smooth_chromatogram(signal, window):
     """smooth-out the chromatogram
-    
+
     :param signal: the chomatogram as 1d array
     :param window: the size of the window
     """
@@ -73,7 +73,7 @@ def smooth_chromatogram(signal, window):
 def search_peaks(signal, wmin=10, scale=0.9):
     """
     Label all peak regions of chromatogram.
-    
+
     :param signal=smooth signal
     :param wmin: minimum width for a peak. smaller ones are discarded.
     :param scale: shrink factor (i.e. <1 for the search zone)
@@ -108,14 +108,14 @@ def search_peaks(signal, wmin=10, scale=0.9):
 def build_background(I, std=None, keep=0.3):
     """
     Build a background from a SVD and search for the frames looking most like the background.
-    
+
     1. build a coarse approximation based on the SVD.
     2. measure the distance (cormap) of every single frame to the fundamental of the SVD
     3. average frames that looks most like the coarse approximation (with deviation)
-    
+
     :param I: 2D array of shape (nframes, nbins)
     :param std: same as I but with the standard deviation.
-    :param keep: fraction of frames to consider for background (<1!), 30% looks like a good guess 
+    :param keep: fraction of frames to consider for background (<1!), 30% looks like a good guess
     :return: (bg_avg, bg_std, indexes), each 1d of size nbins. + the index of the frames to keep
     """
     U, S, V = numpy.linalg.svd(I.T, full_matrices=False)
@@ -135,21 +135,21 @@ def build_background(I, std=None, keep=0.3):
 
 class HPLC(Plugin):
     """ Rebuild the complete chromatogram and perform basic analysis on it.
-    
+
         Typical JSON file:
     {
       "integrated_files": ["img_001.h5", "img_002.h5"],
       "output_file": "hplc.h5"
       "ispyb": {
         "url": "http://ispyb.esrf.fr:1234",
-        "pyarch": "/data/pyarch/mx1234/sample", 
+        "pyarch": "/data/pyarch/mx1234/sample",
         "measurement_id": -1,
         "collection_id": -1
        },
-       "nmf_components": 5, 
+       "nmf_components": 5,
       "wait_for": [jobid_img001, jobid_img002],
       "plugin_name": "bm29.hplc"
-    } 
+    }
     """
     NMF_COMP = 5
     "Default number of Non-negative matrix factorisation components. Correspond to the number of spieces"
@@ -214,7 +214,7 @@ def process(self):
         self.to_pyarch["sample_name"] = self.juices[0].sample.name
         if not self.input.get("no_ispyb"):
             self.send_to_ispyb()
-        # self.output["icat"] = 
+        # self.output["icat"] =
         self.send_to_icat()
 
     def teardown(self):
@@ -233,7 +233,7 @@ def create_nexus(self):
                               title='BioSaxs HPLC experiment',
                               force_time=get_isotime())
         entry_grp["version"] = __version__
-        nxs.h5.attrs["default"] = entry_grp.name
+        nxs.h5.attrs["default"] = entry_grp.name.strip("/")
 
     # Configuration
         cfg_grp = nxs.new_class(entry_grp, "configuration", "NXnote")
@@ -285,7 +285,8 @@ def create_nexus(self):
         frame_ds.attrs["long_name"] = "frame index"
         hplc_data.attrs["signal"] = "sum"
         hplc_data.attrs["axes"] = "frame_ids"
-        chroma_grp.attrs["default"] = entry_grp.attrs["default"] = hplc_data.name
+        chroma_grp.attrs["default"] = posixpath.relpath(hplc_data.name, chroma_grp.name)
+        entry_grp.attrs["default"] = posixpath.relpath(hplc_data.name, entry_grp.name)
         time_ds = hplc_data.create_dataset("timestamps", data=timestamps, dtype=numpy.uint32)
         time_ds.attrs["interpretation"] = "spectrum"
         time_ds.attrs["long_name"] = "Time stamps (s)"
@@ -340,7 +341,7 @@ def create_nexus(self):
         chroma_data.attrs["SILX_style"] = NORMAL_STYLE
 
         svd_grp.create_dataset("eigenvalues", data=S[:r], dtype=numpy.float32)
-        svd_grp.attrs["default"] = chroma_data.name
+        svd_grp.attrs["default"] = posixpath.relpath(chroma_data.name, svd_grp.name)
 
     # Process 3: NMF matrix decomposition
         nmf_grp = nxs.new_class(entry_grp, "3_NMF", "NXprocess")
@@ -370,7 +371,7 @@ def create_nexus(self):
             chroma_ds.attrs["interpretation"] = "spectrum"
             chroma_data.attrs["signal"] = "H"
             chroma_data.attrs["SILX_style"] = NORMAL_STYLE
-            nmf_grp.attrs["default"] = chroma_data.name
+            nmf_grp.attrs["default"] = posixpath.relpath(chroma_data.name, nmf_grp.name)
 
     # Process 5: Background estimation
         bg_grp = nxs.new_class(entry_grp, "4_background", "NXprocess")
@@ -397,7 +398,7 @@ def create_nexus(self):
         bg_q_ds.attrs["long_name"] = f"Scattering vector q ({radius_unit}⁻¹)"
         bg_std_ds = bg_data.create_dataset("errors", data=numpy.ascontiguousarray(bg_std, dtype=numpy.float32))
         bg_std_ds.attrs["interpretation"] = "spectrum"
-        bg_grp.attrs["default"] = bg_data.name
+        bg_grp.attrs["default"] = posixpath.relpath(bg_data.name, bg_grp.name)
         I_sub = I - bg_avg
         Istd_sub = numpy.sqrt(sigma ** 2 + bg_std ** 2)
 
@@ -434,7 +435,7 @@ def one_fraction(self, fraction, index, nxs, top_grp):
         :param index: index of the fraction
         :param nxs: opened Nexus file object
         :param top_grp: top level nexus group to start building into.
-        
+
         """
         q = self.juices[0].q
         unit = self.juices[0].unit
@@ -458,7 +459,7 @@ def one_fraction(self, fraction, index, nxs, top_grp):
         avg_data.attrs["title"] = f"{sample.name}, frames {fraction.start}-{fraction.stop} averaged, buffer subtracted"
         avg_data.attrs["signal"] = "I"
         avg_data.attrs["axes"] = radial_unit
-        f_grp.attrs["default"] = avg_data.name
+        f_grp.attrs["default"] = posixpath.relpath(avg_data.name, f_grp.name)
         avg_q_ds = avg_data.create_dataset(radial_unit,
                                            data=numpy.ascontiguousarray(q, dtype=numpy.float32))
         avg_q_ds.attrs["units"] = unit_name
@@ -610,9 +611,9 @@ def one_fraction(self, fraction, index, nxs, top_grp):
         guinier_data_attrs["signal"] = "logI"
         guinier_data_attrs["axes"] = "q2"
         guinier_data_attrs["auxiliary_signals"] = "fit"
-        guinier_grp.attrs["default"] = guinier_data.name
+        guinier_grp.attrs["default"] = posixpath.relpath(guinier_data.name, guinier_grp.name)
         if guinier is None:
-            f_grp.attrs["default"] = avg_data.name
+            f_grp.attrs["default"] = posixpath.relpath(avg_data.name, f_grp.name)
             self.log_error("No Guinier region found, data of dubious quality", do_raise=False)
             return
 
@@ -625,7 +626,7 @@ def one_fraction(self, fraction, index, nxs, top_grp):
         kratky_data = nxs.new_class(kratky_grp, "results", "NXdata")
         kratky_data.attrs["SILX_style"] = NORMAL_STYLE
         kratky_data.attrs["title"] = "Dimensionless Kratky plots"
-        kratky_grp.attrs["default"] = kratky_data.name
+        kratky_grp.attrs["default"] = posixpath.relpath(kratky_data.name, kratky_grp.name)
 
     # Stage #5 Kratky plot generation:
         Rg = guinier.Rg
@@ -762,7 +763,7 @@ def one_fraction(self, fraction, index, nxs, top_grp):
             bift_ds = avg_data.create_dataset("BIFT", data=T.dot(stats.density_avg).astype(numpy.float32))
             bift_ds.attrs["interpretation"] = "spectrum"
             avg_data.attrs["auxiliary_signals"] = "BIFT"
-            bift_grp.attrs["default"] = bift_data.name
+            bift_grp.attrs["default"] = posixpath.relpath(bift_data.name, bift_grp.name)
 
     def build_ispyb_group(self, nxs, top_grp):
         """Build the ispyb group inside the HDF5/Nexus file and all associated calculation
@@ -942,7 +943,7 @@ def send_to_ispyb(self):
         else:
             self.log_warning(f"Not sending to ISPyB: no valid URL in {self.ispyb}")
 
-    def send_to_icat(self): 
+    def send_to_icat(self):
         to_icat = copy.copy(self.to_pyarch)
         to_icat["experiment_type"] = "hplc"
         to_icat["sample"] = self.juices[0].sample
@@ -954,8 +955,8 @@ def send_to_icat(self):
         return send_icat(sample=self.juices[0].sample,
                          raw=os.path.dirname(os.path.abspath(self.input_files[0])),
                          path=os.path.dirname(os.path.abspath(self.output_file)),
-                         data=to_icat, 
-                         gallery=gallery, 
+                         data=to_icat,
+                         gallery=gallery,
                          metadata=metadata)
 
     def save_csv(self, filename, sum_I, Rg):
@@ -970,7 +971,6 @@ def save_csv(self, filename, sum_I, Rg):
         lines.append("")
         with open(filename, "w") as csv:
             csv.write(os.linesep.join(lines))
-            
-            
-            
-        
\ No newline at end of file
+
+
+
diff --git a/plugins/bm29/integrate.py b/plugins/bm29/integrate.py
index d0e0b2f..c447015 100644
--- a/plugins/bm29/integrate.py
+++ b/plugins/bm29/integrate.py
@@ -11,7 +11,7 @@
 __contact__ = "Jerome.Kieffer@ESRF.eu"
 __license__ = "MIT"
 __copyright__ = "European Synchrotron Radiation Facility, Grenoble, France"
-__date__ = "21/02/2025" 
+__date__ = "04/06/2025"
 __status__ = "development"
 __version__ = "0.3.0"
 
@@ -20,6 +20,7 @@
 import json
 import logging
 import copy
+import posixpath
 from collections import namedtuple
 from urllib3.util import parse_url
 from dahu.plugin import Plugin
@@ -79,6 +80,7 @@ class IntegrateMultiframe(Plugin):
       "fidelity_rel": 1e-3,
       "hplc_mode": 0,
       "timeout": 10,
+      "average_out_monitor_values": False,  # use this to work around noisy beam stop diode reading.
       "sample": {
         "name": "bsa",
         "description": "protein description like Bovine Serum Albumin",
@@ -185,6 +187,8 @@ def setup(self, kwargs=None):
         else:
             self.energy = numpy.float32(self.energy)  # It is important to fix the datatype of the energy
         self.monitor_values = numpy.array(self.input.get("monitor_values", 1), dtype=numpy.float64)
+        if self.input.get("average_out_monitor_values"):
+            self.monitor_values = numpy.zeros_like(self.monitor_values) + self.monitor_values.mean()
         self.normalization_factor = float(self.input.get("normalization_factor", 1))
         self.scale_factor = float(self.input.get("exposure_time", 1)) / self.normalization_factor
 
@@ -232,7 +236,7 @@ def process(self):
         self.create_nexus()
         self.output["memcached"] = self.send_to_memcached()
         self.send_to_ispyb()
-        #self.output["icat"] = 
+        #self.output["icat"] =
         self.send_to_icat()
 
     def wait_file(self, filename, timeout=None):
@@ -277,7 +281,7 @@ def create_nexus(self):
         entry_grp = nxs.new_entry("entry", self.input.get("plugin_name", "dahu"),
                               title='BioSaxs multiframe integration',
                               force_time=get_isotime(creation_time))
-        nxs.h5.attrs["default"] = entry_grp.name
+        nxs.h5.attrs["default"] = entry_grp.name.strip("/")
 
         # Configuration
         cfg_grp = nxs.new_class(entry_grp, "configuration", "NXnote")
@@ -432,11 +436,12 @@ def create_nexus(self):
         hplc_data.attrs["axes"] = "frame_ids"
 
         if self.input.get("hplc_mode"):
-            entry_grp.attrs["default"] = entry_grp.attrs["default"] = integration_grp.attrs["default"] = hplc_data.name
+            entry_grp.attrs["default"] = posixpath.relpath(hplc_data.name, entry_grp.name)
+            integration_grp.attrs["default"] = posixpath.relpath(hplc_data.name, integration_grp.name)
             self.log_warning("HPLC mode detected, stopping after frame per frame integration")
             return
 
-        integration_grp.attrs["default"] = integration_data.name
+        integration_grp.attrs["default"] = posixpath.relpath(integration_data.name, integration_grp.name)
 
     # Process 2: Freesas cormap
         cormap_grp = nxs.new_class(entry_grp, "2_correlation_mapping", "NXprocess")
@@ -466,7 +471,7 @@ def create_nexus(self):
 
         to_merge_ds = cormap_data.create_dataset("to_merge", data=numpy.arange(*cormap_results.tomerge, dtype=numpy.uint16))
         to_merge_ds.attrs["long_name"] = "Index of equivalent frames"
-        cormap_grp.attrs["default"] = cormap_data.name
+        cormap_grp.attrs["default"] = posixpath.relpath(cormap_data.name, cormap_grp.name)
         if self.ispyb.url:
             self.to_pyarch["merged"] = cormap_results.tomerge
 
@@ -497,11 +502,11 @@ def create_nexus(self):
         int_std_ds.attrs["interpretation"] = "image"
         int_std_ds.attrs["formula"] = "sqrt(sum_i(variance_i)/sum_i(normalization_i))"
         int_std_ds.attrs["method"] = "Propagated error from weighted mean assuming poissonian behavour of every data-point"
-        
+
         int_nrm_ds = average_data.create_dataset("normalization", data=norm)
         int_nrm_ds.attrs["formula"] = "sum_i(normalization_i))"
-        
-        average_grp.attrs["default"] = average_data.name
+
+        average_grp.attrs["default"] = posixpath.relpath(average_data.name, average_grp.name)
 
     # Process 4: Azimuthal integration of the time average image
         ai2_grp = nxs.new_class(entry_grp, "4_azimuthal_integration", "NXprocess")
@@ -518,7 +523,7 @@ def create_nexus(self):
         ai2_grp["configuration"] = integration_grp["configuration"]
         # ai2_grp["polarization_factor"] = integration_grp["polarization_factor"]
         # ai2_grp["integration_method"] = integration_grp["integration_method"]
-        ai2_grp.attrs["default"] = ai2_data.name
+        ai2_grp.attrs["default"] = posixpath.relpath(ai2_data.name, ai2_grp.name)
 
     # Stage 4 processing
         intensity_std = res3.deviation
@@ -552,7 +557,7 @@ def create_nexus(self):
         ai2_std_ds.attrs["interpretation"] = "spectrum"
         ai2_int_ds.attrs["units"] = "arbitrary"
         # Finally declare the default entry and default dataset ...
-        entry_grp.attrs["default"] = ai2_data.name
+        entry_grp.attrs["default"] = posixpath.relpath(ai2_data.name, entry_grp.name)
 
         # Export this to the output JSON
         # self.output["q"] = res2.radial
@@ -606,7 +611,7 @@ def process3_average(self, tomerge):
         valid_slice = slice(*tomerge)
         mask = self.ai.detector.mask
         sum_data = (self.input_frames[valid_slice]).sum(axis=0)
-        sum_norm = self.scale_factor * sum(self.monitor_values[valid_slice])    
+        sum_norm = self.scale_factor * sum(self.monitor_values[valid_slice])
         if numexpr is not None:
             # Numexpr is many-times faster than numpy when it comes to element-wise operations
             intensity_avg = numexpr.evaluate("where(mask==0, sum_data/sum_norm, 0.0)")
@@ -629,9 +634,9 @@ def send_to_ispyb(self):
                 self.log_warning(f"Not sending to ISPyB: no valid URL {self.ispyb.url}")
 
     def send_to_icat(self):
-        #Some more metadata for iCat, as strings: 
+        #Some more metadata for iCat, as strings:
         to_icat = copy.copy(self.to_pyarch)
-        to_icat["experiment_type"] = "hplc" if self.input.get("hplc_mode") else "sample-changer"  
+        to_icat["experiment_type"] = "hplc" if self.input.get("hplc_mode") else "sample-changer"
         to_icat["sample"] = self.sample
         to_icat["SAXS_maskFile"] = self.mask
         to_icat["SAXS_waveLength"] = str(self.ai.wavelength)
@@ -645,13 +650,13 @@ def send_to_icat(self):
         f2d = self.ai.getFit2D()
         to_icat["SAXS_beam_center_x"] = str(f2d["centerX"])
         to_icat["SAXS_beam_center_y"] = str(f2d["centerY"])
-        
+
         metadata = {"scanType": "integration"}
         return send_icat(sample=self.sample.name,
                          raw=os.path.dirname(os.path.dirname(os.path.abspath(self.input_file))),
                          path=os.path.dirname(os.path.abspath(self.output_file)),
-                         data=to_icat, 
-                         gallery=self.ispyb.gallery or os.path.join(os.path.dirname(os.path.abspath(self.output_file)), "gallery"), 
+                         data=to_icat,
+                         gallery=self.ispyb.gallery or os.path.join(os.path.dirname(os.path.abspath(self.output_file)), "gallery"),
                          metadata=metadata)
 
     def send_to_memcached(self):
@@ -661,5 +666,5 @@ def send_to_memcached(self):
         for k in sorted(self.to_memcached.keys(), key=lambda i:self.to_memcached[i].nbytes):
             key = f"{key_base}_{k}"
             dico[key] = json.dumps(self.to_memcached[k], cls=NumpyEncoder)
-        return to_memcached(dico) 
+        return to_memcached(dico)
 
diff --git a/plugins/bm29/nexus.py b/plugins/bm29/nexus.py
index b8d214a..411f949 100644
--- a/plugins/bm29/nexus.py
+++ b/plugins/bm29/nexus.py
@@ -4,7 +4,7 @@
 __contact__ = "Jerome.Kieffer@ESRF.eu"
 __license__ = "MIT"
 __copyright__ = "European Synchrotron Radiation Facility, Grenoble, France"
-__date__ = "21/02/2025"
+__date__ = "27/05/2025"
 __status__ = "production"
 __docformat__ = 'restructuredtext'
 
@@ -85,8 +85,8 @@ class Nexus:
     TODO: make it thread-safe !!!
     """
 
-    def __init__(self, filename, mode=None, 
-                 creator=None, 
+    def __init__(self, filename, mode=None,
+                 creator=None,
                  timeout=None,
                  start_time=None):
         """
@@ -244,7 +244,7 @@ def new_entry(self, entry="entry", program_name="pyFAI",
             nb_entries = len(self.get_entries())
             entry = "%s_%04i" % (entry, nb_entries)
         entry_grp = self.h5.require_group(entry)
-        self.h5.attrs["default"] = entry
+        self.h5.attrs["default"] = entry_grp.name.strip("/")
         entry_grp.attrs["NX_class"] = "NXentry"
         entry_grp["title"] = str(title)
         entry_grp["program_name"] = program_name
@@ -269,7 +269,7 @@ def new_instrument(self, entry="entry", instrument_name="id00",):
     def new_class(grp, name, class_type="NXcollection"):
         """
         create a new sub-group with  type class_type
-        
+
         :param grp: parent group
         :param name: name of the sub-group
         :param class_type: NeXus class name
diff --git a/plugins/bm29/subtracte.py b/plugins/bm29/subtracte.py
index 75246df..4ea0cbe 100644
--- a/plugins/bm29/subtracte.py
+++ b/plugins/bm29/subtracte.py
@@ -3,17 +3,17 @@
 
 """Data Analysis plugin for BM29: BioSaxs
 
-* SubtractBuffer: Search for the equivalence of buffers, average them and subtract from sample signal.  
-* SaxsAnalysis: Performs Guinier + Kratky + IFT, generates plots  
+* SubtractBuffer: Search for the equivalence of buffers, average them and subtract from sample signal.
+* SaxsAnalysis: Performs Guinier + Kratky + IFT, generates plots
 """
 
 __authors__ = ["Jérôme Kieffer"]
 __contact__ = "Jerome.Kieffer@ESRF.eu"
 __license__ = "MIT"
 __copyright__ = "European Synchrotron Radiation Facility, Grenoble, France"
-__date__ = "24/02/2025"
+__date__ = "27/05/2025"
 __status__ = "development"
-__version__ = "0.3.0" 
+__version__ = "0.3.0"
 
 import os
 import posixpath
@@ -141,7 +141,7 @@ def teardown(self):
         self.output["Dmax"] = self.Dmax
         self.output["Vc"] = self.Vc
         self.output["mass"] = self.mass
-        
+
         #teardown everything else:
         if self.nxs is not None:
             self.nxs.close()
@@ -169,7 +169,7 @@ def process(self):
                 raise(err)
         else:
             self.send_to_ispyb()
-            self.send_to_icat()        
+            self.send_to_icat()
         self.output["memcached"] = self.send_to_memcached()
 
 
@@ -209,7 +209,7 @@ def create_nexus(self):
         entry_grp = nxs.new_entry("entry", self.input.get("plugin_name", "dahu"),
                                   title='BioSaxs buffer subtraction',
                                   force_time=get_isotime())
-        nxs.h5.attrs["default"] = entry_grp.name
+        nxs.h5.attrs["default"] = entry_grp.name.strip["/"]
 
     # Configuration
         cfg_grp = nxs.new_class(entry_grp, "configuration", "NXnote")
@@ -271,7 +271,7 @@ def create_nexus(self):
         to_merge_ds = cormap_data.create_dataset("to_merge", data=to_merge_idx)
         # self.log_warning(f"to_merge: {tomerge}")
         to_merge_ds.attrs["long_name"] = "Index of equivalent frames"
-        cormap_grp.attrs["default"] = cormap_data.name
+        cormap_grp.attrs["default"] = posixpath.relpath(cormap_data.name, cormap_grp.name)
 
     # Process 2: Image processing: subtraction with standard deviation
         average_grp = nxs.new_class(entry_grp, "2_buffer_subtraction", "NXprocess")
@@ -283,7 +283,7 @@ def create_nexus(self):
         average_data.attrs["signal"] = "intensity_normed"
     # Stage 2 processing
 
-        # Nota: This formula takes into account the number of input frames in each averaged buffer !        
+        # Nota: This formula takes into account the number of input frames in each averaged buffer !
         #  avg = Σdata / Σnorm
         #  var = sigma² = ΣV / Σnorm
         # TODO implement those math using numexpr:
@@ -310,7 +310,7 @@ def create_nexus(self):
         int_std_ds.attrs["interpretation"] = "image"
         int_std_ds.attrs["formula"] = "sqrt( sample_variance + weighted_mean(buffer_variance_i) )"
         int_std_ds.attrs["method"] = "quadratic sum of sample error and buffer errors"
-        average_grp.attrs["default"] = average_data.name
+        average_grp.attrs["default"] = posixpath.relpath(average_data.name, average_grp.name)
 
         key_cache = KeyCache(self.sample_juice.npt, self.sample_juice.unit, self.sample_juice.poni, self.sample_juice.mask, self.sample_juice.energy)
         ai = get_integrator(key_cache)
@@ -346,7 +346,7 @@ def create_nexus(self):
         ai2_data.attrs["title"] = "%s, subtracted" % self.sample_juice.sample.name
         ai2_data.attrs["signal"] = "I"
         ai2_data.attrs["axes"] = radial_unit
-        ai2_grp.attrs["default"] = ai2_data.name
+        ai2_grp.attrs["default"] = posixpath.relpath(ai2_data.name, ai2_grp.name)
         cfg_grp = nxs.new_class(ai2_grp, "configuration", "NXnote")
         cfg_grp.create_dataset("data", data=json.dumps(ai.get_config(), indent=2, separators=(",\r\n", ": ")))
         cfg_grp.create_dataset("format", data="text/json")
@@ -393,7 +393,7 @@ def create_nexus(self):
 
         #  Finally declare the default entry and default dataset ...
         #  overlay the BIFT fitted data on top of the scattering curve
-        entry_grp.attrs["default"] = ai2_data.name
+        entry_grp.attrs["default"] = posixpath.relpath(ai2_data.name, entry_grp.name)
 
     # Process 4: Guinier analysis
         guinier_grp = nxs.new_class(entry_grp, "4_Guinier_analysis", "NXprocess")
@@ -527,9 +527,9 @@ def create_nexus(self):
         guinier_data_attrs["signal"] = "logI"
         guinier_data_attrs["axes"] = "q2"
         guinier_data_attrs["auxiliary_signals"] = "fit"
-        guinier_grp.attrs["default"] = guinier_data.name
+        guinier_grp.attrs["default"] = posixpath.relpath(guinier_data.name, guinier_grp.name)
         if guinier is None:
-            entry_grp.attrs["default"] = ai2_data.name
+            entry_grp.attrs["default"] = posixpath.relpath(ai2_data.name, entry_grp.name)
             self.log_error("No Guinier region found, data of dubious quality", do_raise=True)
 
     # Process 5: Kratky plot
@@ -541,7 +541,7 @@ def create_nexus(self):
         kratky_data = nxs.new_class(kratky_grp, "results", "NXdata")
         kratky_data.attrs["SILX_style"] = NORMAL_STYLE
         kratky_data.attrs["title"] = "Dimensionless Kratky plots"
-        kratky_grp.attrs["default"] = kratky_data.name
+        kratky_grp.attrs["default"] = posixpath.relpath(kratky_data.name, kratky_grp.name)
 
     # Stage #5 Kratky plot generation:
         Rg = guinier.Rg
@@ -552,9 +552,9 @@ def create_nexus(self):
         qRg_ds = kratky_data.create_dataset("qRg", data=xdata.astype(numpy.float32))
         qRg_ds.attrs["interpretation"] = "spectrum"
         qRg_ds.attrs["long_name"] = "q·Rg (unit-less)"
-        
+
         #Nota the "/" hereafter is chr(8725), the division sign and not the usual slash
-        k_ds = kratky_data.create_dataset("q2Rg2I∕I0", data=ydata.astype(numpy.float32)) 
+        k_ds = kratky_data.create_dataset("q2Rg2I∕I0", data=ydata.astype(numpy.float32))
         k_ds.attrs["interpretation"] = "spectrum"
         k_ds.attrs["long_name"] = "q²Rg²I(q)/I₀"
         ke_ds = kratky_data.create_dataset("errors", data=dy.astype(numpy.float32))
@@ -681,7 +681,7 @@ def create_nexus(self):
             bift_ds = ai2_data.create_dataset("BIFT", data=T.dot(stats.density_avg).astype(numpy.float32))
             bift_ds.attrs["interpretation"] = "spectrum"
             ai2_data.attrs["auxiliary_signals"] = "BIFT"
-            bift_grp.attrs["default"] = bift_data.name
+            bift_grp.attrs["default"] = posixpath.relpath(bift_data.name, bift_grp.name)
             self.to_pyarch["bift"] = stats
 
     @staticmethod
@@ -739,7 +739,7 @@ def send_to_ispyb(self):
         else:
             self.log_warning("Not sending to ISPyB: no valid URL %s" % self.ispyb.url)
 
-    def send_to_icat(self): 
+    def send_to_icat(self):
         to_icat = copy.copy(self.to_pyarch)
         to_icat["experiment_type"] = "sample-changer"
         if self.sample_juice is None:
@@ -752,8 +752,8 @@ def send_to_icat(self):
         return send_icat(sample=self.sample_juice.sample,
                          raw=raw,
                          path=os.path.dirname(os.path.abspath(self.output_file)),
-                         data=to_icat, 
-                         gallery=self.ispyb.gallery or os.path.join(os.path.dirname(os.path.abspath(self.output_file)), "gallery"), 
+                         data=to_icat,
+                         gallery=self.ispyb.gallery or os.path.join(os.path.dirname(os.path.abspath(self.output_file)), "gallery"),
                          metadata=metadata)
 
     def send_to_memcached(self):
@@ -764,5 +764,5 @@ def send_to_memcached(self):
             key = key_base + "_" + k
             dico[key] = json.dumps(self.to_memcached[k], cls=NumpyEncoder)
 
-        return to_memcached(dico) 
+        return to_memcached(dico)
 
diff --git a/plugins/id02/nexus.py b/plugins/id02/nexus.py
index 3978bf3..a27296f 100644
--- a/plugins/id02/nexus.py
+++ b/plugins/id02/nexus.py
@@ -4,7 +4,7 @@
 __contact__ = "Jerome.Kieffer@ESRF.eu"
 __license__ = "MIT"
 __copyright__ = "European Synchrotron Radiation Facility, Grenoble, France"
-__date__ = "20/02/2025"
+__date__ = "27/05/2025"
 __status__ = "production"
 __docformat__ = 'restructuredtext'
 
@@ -86,8 +86,8 @@ class Nexus:
     TODO: make it thread-safe !!!
     """
 
-    def __init__(self, filename, mode=None, 
-                 creator=None, 
+    def __init__(self, filename, mode=None,
+                 creator=None,
                  timeout=None,
                  start_time=None):
         """
@@ -233,7 +233,7 @@ def new_entry(self, entry="entry", program_name="pyFAI",
             nb_entries = len(self.get_entries())
             entry = "%s_%04i" % (entry, nb_entries)
         entry_grp = self.h5.require_group(entry)
-        self.h5.attrs["default"] = entry
+        self.h5.attrs["default"] = entry.strip("/")
         entry_grp.attrs["NX_class"] = "NXentry"
         entry_grp["title"] = str(title)
         entry_grp["program_name"] = program_name
@@ -258,7 +258,7 @@ def new_instrument(self, entry="entry", instrument_name="id00",):
     def new_class(grp, name, class_type="NXcollection"):
         """
         create a new sub-group with  type class_type
-        
+
         :param grp: parent group
         :param name: name of the sub-group
         :param class_type: NeXus class name
diff --git a/plugins/id02/single_detector.py b/plugins/id02/single_detector.py
index 33dcb1f..0c87450 100644
--- a/plugins/id02/single_detector.py
+++ b/plugins/id02/single_detector.py
@@ -8,7 +8,7 @@
 __contact__ = "Jerome.Kieffer@ESRF.eu"
 __license__ = "MIT"
 __copyright__ = "European Synchrotron Radiation Facility, Grenoble, France"
-__date__ = "20/02/2025"
+__date__ = "27/05/2025"
 __status__ = "development"
 __version__ = "0.9.3"
 
@@ -683,7 +683,7 @@ def create_hdf5(self):
             nxprocess["date"] = isotime
             nxprocess["processing_type"] = ensure_str(ext)
             nxdata = nxs.new_class(nxprocess, "result_" + ext, class_type="NXdata")
-            entry.attrs["default"] = nxdata.name
+            entry.attrs["default"] = posixpath.relpath(nxdata.name, entry.name)
             metadata_grp = nxprocess.require_group("parameters")
 
             for key, val in self.metadata.items():
@@ -870,7 +870,7 @@ def grpdeepcopy(name, obj):
                                               **compression)
             nxdata.attrs["signal"] = "data"
             # output_ds.attrs["signal"] = "1"
-            entry.attrs["default"] = nxdata.name
+            entry.attrs["default"] = posixpath.relpath(nxdata.name, entry.name)
             if self.variance_formula is not None:
                 error_ds = nxdata.create_dataset("data_errors", shape,
                                                  dtype=numpy.float32,
diff --git a/plugins/id02/xpcs.py b/plugins/id02/xpcs.py
index 2fcbed3..5979500 100644
--- a/plugins/id02/xpcs.py
+++ b/plugins/id02/xpcs.py
@@ -1,5 +1,5 @@
 """
-X-ray photon correlation spectroscopy plugin for ID02 
+X-ray photon correlation spectroscopy plugin for ID02
 
 """
 
@@ -7,12 +7,13 @@
 __contact__ = "Jerome.Kieffer@ESRF.eu"
 __license__ = "MIT"
 __copyright__ = "European Synchrotron Radiation Facility, Grenoble, France"
-__date__ = "20/02/2025"
+__date__ = "27/05/2025"
 __status__ = "development"
 __version__ = "0.1.1"
 
 import os
 import json
+import posixpath
 import logging
 logger = logging.getLogger("id02.xpcs")
 
@@ -41,7 +42,7 @@
 
 
 class XPCS(Plugin):
-    """This plugin does pixel correlation for XPCS and averages the signal from various bins provided in the qmask. 
+    """This plugin does pixel correlation for XPCS and averages the signal from various bins provided in the qmask.
 
 Minimalistic example:
 {
@@ -70,13 +71,13 @@ class XPCS(Plugin):
         "q_mask": "qmask.npy",
         "beamstop_mask": "mask.npy" ,
         "directbeam_x": 104, #pixel
-        "directbeam_y": 157, #pixel 
+        "directbeam_y": 157, #pixel
     },
     "correlator":{
         "name": "MatMulCorrelator",
         "dtype": "uint8"
     }
-             
+
 }
 """
 
@@ -251,7 +252,7 @@ def save_results(self, result):
                                       program_name=self.input.get("plugin_name", "dahu"),
                                       title="XPCS experiment",
                                       force_time=self.start_time)
-            nxs.h5.attrs["default"] = entry_grp.name
+            nxs.h5.attrs["default"] = entry_grp.name.strip("/")
 
             # Sample description, provided by the input
             sample_grp = nxs.new_class(entry_grp, "sample", "NXsample")
@@ -310,7 +311,8 @@ def save_results(self, result):
             qmask_ds.attrs["interpretation"] = "image"
             qmask_ds.attrs["long_name"] = "mask with bins averaged (0=masked-out)"
 
-            entry_grp.attrs["default"] = xpcs_grp.attrs["default"] = xpcs_data.name
+            entry_grp.attrs["default"] = posixpath.relpath(xpcs_data.name, entry_grp.name)
+            xpcs_grp.attrs["default"] = posixpath.relpath(xpcs_data.name, xpcs_grp.name)
             result_ds = xpcs_data.create_dataset("g2", data=result.res, chunks=True, **COMPRESSION)
             result_ds.attrs["interpretation"] = "spectrum"
             errors_ds = xpcs_data.create_dataset("errors", data=result.dev, chunks=True, **COMPRESSION)