neumoDVB

[Llew]

.....and how do I go about that? I'm a new at using Linux.

The same way you are installing your dependencies. For Synaptic,

sudo apt install -y synaptic

 

[deeptho]

A new version of neumoDVB will be available soon. I am working on fixing some new
bugs, adding some documentation of new features and cleaning the code a bit.

One of the changes is that scanning muxes now has more options:

• You can choose whether or not to scan the full EPG. Scanning full EPG is important for viewing, but while DX-ing you probably prefer scanning speed.
• If new muxes, which you did not request to scan, are discovered then you can choose whether or not to scan them as well.
• You can choose whether to use blind tune during scan, and allow or prevent the dish
• to be moved
• You can restrict the scan to specific dishes (e.g., when you know some are not big enough)
• You can also restrict the scan to specific cards

While all of this can be specified, the defaults are usually fine. neumoDVB will use as many resources as possible to scan as fast as possible.

The code is not yet released.

Other change relate to fully blindscanning all satellites, again with all kinds of options. This is 95% complete. There are all kinds of fixes as well. After the next release, I plan to add
"scan timers" which will periodically do all of the above without human intervention. At that
stage neumoDVB can be used as a replacement for kodi+tvheadend. In some respects
neumoDVB still lags behind for that purpose, but it will have some advantages as well:

• reliable timeshift
• jumping forward and back without spinning circles
• recording a program already in progress (from the time you started viewing)
• no weird problems on some channels

 

[Babadem]

I'm stuck Here

Now install the blindscan drivers:

cd blindscan_kernel
git clone --depth=1 deeptho/linux_media -b deepthought ./media

When I type the commands above, I get the following errors below:

git clone --depth=1 deeptho/linux_media -b deepthought ./media
bash: cd: blindscan_kernel: No such file or directory
fatal: repository 'deeptho/linux_media' does not exist

 

[Llew]

I'm stuck Here

Now install the blindscan drivers:




When I type the commands above, I get the following errors below:

Seems you've missed making the directory.

Scroll down to the installation instructions in GitHub - deeptho/linux_media: neumoDVB blindscan drivers media_tree

 

[ptergrand]

It would be better if the monitor fits.

 

[Babadem]

Seems you've missed making the directory.

Scroll down to the installation instructions in GitHub - deeptho/linux_media: neumoDVB blindscan drivers media_tree

I seem to be running around in circles. I'm temporarily giving this a rest for now.

 

[Llew]

I seem to be running around in circles. I'm temporarily giving this a rest for now.

If you explain to us the part you are having difficulty with, I'm sure one of us can give you some help.

 

[ptergrand]

When will the new version be released?

 

[deeptho]

When will the new version be released?

Soon. Just fixing some bugs. Found a big one today. I also need to update documentation and change list

 

[deeptho]

Here are a few more screenshots illustrating a new feature: it is now also possible to scan multiple satellites from the "satellites" list,
while restricting the scan to specific polarisations or frequency bands, and this without bringing up the spectrum dialog. The main advantage
is that this simplifies scanning many satellites. In a future edition, it will also be possible to run such a scan fully automatically, for example,
to rescan periodically from a timer.


The above screenshots whow what is possible. After scanning, the satellite list will also record the data of the most recent spectrum
acquisition:

 

[Llew]

Splendid deeptho. That's going to make things easier to see at a glance what we are using to scan the satellite(s) chosen, and the results found, all in convenient charts.

No doubt there will be tick boxes for other bands too in the Allowed Bands section when the new version is available.

 

[ptergrand]

You are amazing.

 

[deeptho]

Splendid deeptho. That's going to make things easier to see at a glance what we are using to scan the satellite(s) chosen, and the results found, all in convenient charts.

No doubt there will be tick boxes for other bands too in the Allowed Bands section when the new version is available.

The list of bands is defined from what is available on the selected satellites. For example KaA wil be shown if a KaA sat has been selected. You can always add sats that are missing.

 

[RyLe]

Hi

The latest Fedora39 kernel (6.6.2-201.fc39.x86_64) broke blindscan kernel compilation with too many v4l internal structures modifications for me to simply patch like I did the previous times.

The last Fedora 39 kernel that works with all of my previous relatively simple modifications is the 6.5.12-300.fc39.x86_64.

Deeptho, do you plan to support the 6.6 kernel branch with the upcoming neumodvb update ?

 

[deeptho]

Here is a list of the changes in the next release - Part 1

Changes in version neumodvb-1.4​

Main changes​

• Allow scanning all existing muxes on multiple satellites, while selecting muxes based on polarisation and band.
• Allow scanning spectral bands on multiple satellites, while selecting bands based on polarisation and band and on frequency range.
• Have muxinfo display more info during scanning.
• Satellite list now shows information on last band scan in a multi-row format.
• Improved code for handling conflicts during subscriptions.
• Allow mux and service subscription to share same active mux.
• Various bug fixes related to incorrect selection of LNBs or not powering up LNBs or selecting bands when multples muxes are tuned in parallel. This also leads to more reliable scanning.
• Refactor scan code to make it less complicated and easier to maintain.
• Fixes for recovering recordings when neumoDVB is restarted while a recording is in progress.
• Add autorec code, which checks all new EPG data and creates recording records based on matching program name, story, start time, end time…
• sat_list is now organized per frequency band dvbc and dvbt are no longer included in that list. Having a different satellite entry per frequency band makes it easier to display only those satellites supporting a specific band.
• Use ch_order=65535 for new services so that they appear at end of service list and display 65535 as empty string. The end result is that for services without a ch_order are now displayed with a blank channel number and at the end of the list, as they are often not of interest to the user.
• Distinguish between cur_lnb_pos (what position does lnb point to) and cur_sat_pos (what as the last network used on lnb). The difference is important when multiple closely spaced satellites are received by the same lnb.
• Channel epg list has been improved: it allows filtering for a specifuic service. The info window has been removed. Instead, the story is shown using multiple lines in a separate column.
• Much faster Sky UK and Freesat EPG processing due to various improvements to internal huffman decoders. All Sky UK epg can now be grabbed in less than 30 seconds.
• The EPG code, and some other code now also runs in parallel when scanning multiple muxes at once. The lmdb database still leads to some serialization.
• All debugging code now uses libfmt, leading to faster logging and more informative log messages with cleaner code.
• Internal database code has been improved to allow parallel calling from multiple threads without unneeded committing of transactions.
• Muxes now use the new mux_id as part of new primary for mux. This makes it easier to avoid bugs causing the wrong mux record to be updated or deleted, when multiple muxes overlap.
• In the service list the string column mux_desc has now been replaced with separate frequency and polarisation columns, allowing filtering and sorting by frequency or polarisation.
• The database now also stores schema_version in database records and in data structures, to be able to detect major schema upgrades.

Spectrum scan and positioner dialog​

• Send lnb error to tune_mux_panel rather than handle it via global subscriber.
• Improved error message when lnb action fails.
• When tune_mux_panel action fails, handle resulting error in spectrum_dialog or positioner_dialog.
• Reset get spectrum button if spectrum scan fails.
• Bug: spectra acquired within one minute not treated as the same in GUI, resulting in hiding of first spectrum when second is displayed.
• Move start_time from spectrum_options to internal fe state.
• Make get_spectrum also return peak and freq data to scan notification code.
• Bug: Indexing of spectra no longer correct.

More reliable tuning and preventing conflicts between parallel tunes​

• Incorrect comparison when no service was reserved.
• New fe_subscribe code.
• Report error message when lnb cannot be subscribed exclusively.
• Bug: mux list popup in positioner dialog shows Ka band muxes when LNB is of Ku type.
• Bug: frontend info not retrieved for adapters in use and file descriptor not closed.
• Bug: Incorrect live_service processing causes deletion of live buffers in use.
• Faster channel changing: instead of first calling unsubscribe from neumompv and then resubscribing, let tuner code do the unsubscribing. This can often reuse an existing active adapter and avoids closing and then reopening a frontend.
• Bug: active_adapter released when unsubscribing one of multiple subscriptions.
• Bug: cannot call start_running in constructor of active_adapter.
• Identify live_service_t by owner and subscription_id; remove epg field in live_service_t to simplify code.
• Incorrect handling of lock loss.
• Confusion between ret and errno.
• Bug: incorrect handling of resource conflicts.
• Bug: incorrect handling of unsubscription when multiple subscriptions use same fe.
• Bug: when subscription fails, active_adapter and subscription are not released.
• Bug: incorrect sending diseqc messages and changing voltage/pol during retune.
• Bug: active_adapter recreated even when adapter remains the same. New fe_subscribe code.
• Bug: assertion when tuning dvbt mux.
• Increase sleep times after changing voltage or rf input.
• Bug: when switching to a different lnb on a frontend, we need to allow for sufficient power up time to prevent tuning from failing.
• Do not power down tuner voltage before switching to a new lnb. Instead start with the old voltage (faster and better).
• Bug: diseqc commands not sent when switching to the same lnb via another card or rf_input (e.g., to scan another polarisation).
• Also send voltage commands afer rf_input has changed; otherwise tuning may fail.
• Bug: not checking if subscription uses same sat/pol/band when checking if frontend in use can be used leads to reusing subscription from another lnb.
• Bug: when reusing the same adapter, tuning uses old lnb.
• Bug: incorrect decision that card cannot be used.
• Bug: incorrect detection of need_diseqc.
• Bug: sat_pos not taken into account when checking if subscription can be reused.
• Bug: subscribe not handling properly the case where a subscription subscribes to a service (or to the mux itself) on the same mux as its old subscription.
• Unify subscribe_mux and matching_existing_subscription code for all dvb types.
• Bug: reserve_fe_in_use not storing mux_key in subscription when subscribing to mux (instead of service).
• Simplify subscriber notifications.
• Allow non exclusive satellite band reservations.
• drop may_control_lnb, may_move_dish, need_blind_tune, need_spectrum and loc a function arguments; use tune_options instead. Also remove them from subscription_ret_t. Remove tune_pars_t and replace with tune_optons_t.
• move several function parameters into tune_options_t; specifically include usals_location, dish_move_penalty and resource_reuse_bonus in tune_options.
• Switch most “subscribe” calls to using tune_options_t parameter.
• Bug: deadlock when reading last_signal_info.
• Incorrect debug message when overriding obviously incorrect si data with driver data.
• Simplify update_mux and its callbacks.
• constness of merge_muxes argument.

Improved mux and satellite band scanning​

• Add allowed_rf_paths, allowed_dishes, and allowed_card_mac_addresses to tune_options to more finely control what resources can be used in tuning.
• Last blind scan setting in positioner and spectrum dialog now remembered as future default.
• Bug: initial scan_report not shown.
• Improved decisions on whether to use mux data from database while scanning spectral peaks.
• Allow add_spectral_peaks to be also called with data from fe_monitor and handle the way that received spectra asre processed: send them also to scanner_t code.
• Bug: incorrect scan statistics.
• Bug: ensure that canceling subscription correctly ends scan.
• Bug: mux_common and scan_id lost when rescanning peak.
• Bug: incorrect rescanning on scan_result_t:OTS.
• Gather scan_stats at point of change directly instead of returning them from function calls.
• Bug: incorrect detection of subscribing to a frequency peak.
• No longer include scan_stats in scan_report. Rename scan_report to scan mux_end_report_t. Ensure that notify_scan_mux_end is also called for muxes that fail to tune at subscription time. Refactor scan_loop.
• Separate code for scanning peaks and muxes.
• Make all notification functions thread safe, while still notifying python layers without delay whenever possible. Harmonize naming of notification functions.
• Safe deletion in ~scanner_t.
• Bug: when unsubscribing, data stored in fe_t and in database becomes inconsistent, leading to never ending scan.
• Bug: incorrect handling of incomplete BAT tables.
• Ignore DefaultAuthorityDescriptorTag to remove many log messages.
• Bug: incorrect frequency from si data overwrites correct data from driver.
• Bug: when reusing active_adapter old fds remain open, leading to incorrect early notification of scanner.
• Bug: last_signal_info and setting lock status in frontend.cc was performed in two steps, leading tuner_thread to see inconsistent information (is_dvb not yet set when lock is reported).
• Bug: incorrect reuse of existing subscription when scanning spectral peaks.
• Ensure generated mux_id always > 0.
• Bug: incorrect growing of vectors of plain old data. Leads to bouquet processing incorrectly expiring entries on 28.2E.
• Bug: when reusing active_adapter old fds remain open, leading to incorrect early notification of scanner.
• Ignore matype for dvbc and dvbt

 

[deeptho]

Here is a list of the changes in the next release - Part 2

Recording​

• Bug: recording not recovered from live recordings at restart.
• Bug: when crashing multiple times, second part of ongoing recording overwrites earlier recovered part.
• Add autorec and code for checking autorecs; Add autoreclist and add autorec dialog on various screens.
• Add make_unique_id for recordings.
• Bug: service.find_by_key not working.
• Bug: autorec_id not correctly computed when autorelist is empty
• Bug: autorec list not updated after adding autorec.
• Bug: reclist not updated after adding recording.
• Ignore database versions in recordings so that old recordings still play back. This relies on some structures remaining valid.
• Added owner field to rec_t. Start new tuner_thread for each active adapter.
• Move recording code from active_adapter to tuner_thread or recmgr_thread.
• Clean recording status of epg records, which may not be uptodate after a crash.

GUI improvements and changes​

• Do not show filter menu when right clicking on icons column.
• Improved display of service in frontend list.
• Ensure that satellite list is created when starting with empty database.
• CmdNew and CmdDelete should not be enabled on live panel.
• Allow selecting multiple non-contiguous rows in lists by ctrl-click.
• Add substring matches for string filtering and make it default.
• Display subscribed services in frontend list.
• Improved display of subscriptions in frontendlist.
• Display shorter matype str.
• Try to preserve position on screen of currently selected item.
• More logical sort order for dates/times.
• Receiver may only exit after all application windows have been closed.
• Font too big in chg selector on chgm screen.
• chglist: use only 1 set of field_matchers.
• Bug: assertion when selecting DVB type in frontend list.
• Bug: incorrect display of services on live screen if some service changes in the background.
• Bug: incorrect sat_pos and usals_pos set after editing lnb.
• Bug in channel group member list.
• Bug: speak no longer working.
• Bug: group select text font cut off on live panel.
• Bug: incorrect display of various fields in autoreclist.
• Bug: autorec_dialog code was incomplete: values not updated when saving.
• Bug: subtitle language selection not working.
• Bug: screen update ignores filter.
• Bug: incorrect handling of control parameters in dvb text strings

EPG improvements​

• New channel epg list: no more infow, allow selecting services, allow showing all services, show story on multiple lines.
• Add service_name to epg records. Also add full service_key to epg records.
• Bug: due to libiconv problem, incorrect handling of larger than usual epg stories.
• Bug: sky epg scan never ends when summary data has no corresponding title.
• clear parser_status when receiving section with bad crc
• Use system tz db rather than downloaded one. Speeds startup of epg screen
• Distinguish between epg and nit scanning in statistics
• epg_scan completeness wrongly computed
• Avoid assertion on incorrect epg data with last_table_id smaller than first_table_id
• Fix channel_epg
• Implemented much faster opentv huffman decoder resulting in 2.5x faster opentv string decoder.
• Bug: gap between epg records sometimes shown as part of older epg record in grid epg
• Show also epg data from the recent past
• Remove epg records in live buffers and recordings (epg is still stored in rec_t record)
• Handle case where epg_screen is None
• Increase number of epg records per section callback to avoid reallocation.
• Incorrect assertion, leading to chrash when processing sky epg.
• Allow filtering in chepg_list when specific service is selected

Compiling, debugging and installing​

• Switch to libfmt fort logging, resulting in faster and more versatile code. Removed xformat and ssaccu.
• Reduce log verbosity.
• Add recmgr to logger output.
• Improved logging.
• Avoid accumulating error messages
• Make stackstring functions gdb friendly again.
• Provide fmt interface for stackstring.
• Provide fmt interfaces for pts_dts_t,m pcr_t and millisonds_t.
• Removed sprintf-like functions in stackstring Removed dtdebugx, dterrox, …
• Remove all operator<< based debug and error calls. Remove dtdebug, dterror, dtinfo, …
• Avoid cgdb crash: disable vector printing.
• Improved ndc in log messages (always show tuner number)
• Do not compile unused code
• Fix incorrect assertions.
• Bug: Invalid memory access.
• Suppress cmake warning
• Correctly indicate which symbols are exported by libneumoreceiver.so
• Default optimisation
• Replace date library (removing slowness of current_timezone) by std::chrono and libfmt
• Fedora38 installation instructions
• remove dead code.
• stackstring: conditionally disable asserts.
• Added ubuntu 23.04 installation instructions
• Compilation fixes for fedora38

Database code improvements​

• New database access code, allowing multithreaded sharing of wtxn (tested and working on epgdb, still single thread) cursors: add readonly field.
• Ignore database versions in recordings so that old recordings still play back. This relies on some structures remaining valid.
• New database schema for muxes and services: introduce mux_id as part of new primary for mux and remove extra_id. Add mux_id column in service list and mux list. Also, replace mux_desc with frequency and polarisation in services, allowing filtering of services based on frequency or polarisation. Add schema_version in database records and in data structures, to be able to detect major schema upgrades.
• Code for handling major database upgrades from python and specifically converting to the database format needed for neumodvb-1.3.
• Improved logging of auto upgrade.
• Allow receiver initialization to be interrupted for major database. upgrade and then for initialization to continue.
• Compute upgrade_dir in python and pass to c++ code.
• Introduce update_record_at_cursor.
• Allow cursor to be explicitly closed before their object is destroyed, this allowing transaction to be aborted.
• Add uuid in schema.
• Improve neumodbstats.py.
• Remove fake satellites with sat_pos_dvbc and sat_pos_dvbt satellite position.
• Improved checking of bad strings in deserialisation code.
• Make testmpm.py work again.
• Bug: deserialize_field_safe: correctly skip variable size fields in old record.
• Bug: cursors are not properly destroyed if no record is read with them.
• Bug: txnmgr accidentally removes open transaction.

Multithreading and task scheduling​

• A new thread is now created for each tuner. This allows parallel processing of EPG on multiple muxes.
• Identify multiple tuner threads by adapter_no.
• Add thread name for recmgr.
• Allow task_queue_t::cb() to work even when thread has exited.
• Disallow pushing tasks after exit() has been called.
• Tuner: run_taks executed too many times.
• BUG: various bugs in txnmgr.
• Bug: Incorrect releasing of wtxn in txnmgr when called from multiple threads.
• Bug: Incorrect locking.
• Bug: accessing released memory when waiting on a future to call thread end.
• Bug: iconv code in stackstring not threadsafe.
• Bug: tasks for unsubscribing were removed.
• Bug: update_lnb called from wrong thread
• Bug: pmt_parser destroys its own fiber at exit, causing bus error

Various improvements and changes​

• Add variant for service and band_scan in subscription_data to be able to store scan_id for band_scans. Could be done with service as well.
• Fix variant types in database.
• Add adapter_no to spectrum file name.
• Introduce tune_options.tune_pars to gather restrictions imposed by sharing resources with other subscriptions.
• Add to_str for enums in database generator. Distinghuish between various satellite bands in scan code.
• Avoid assertion in startup code.
• Raise assertion when using fefd which is not open.
• Bug: division by zero.
• Stackstring: avoid calling constructors when copying vectors containing trivial types. Leads to large. speedup in debug build. Speeds up operator[] as well. Bug: size of zero terminated string.
• Bug in growing memory of stackstring. Remove grow function. inline more stackstring code. Various optimizations.
• hide crc_is_correct and make inline.
• Replace map with array in code managing section flagging, resulting in speedup.
• Additional inlining; suppress warnings when compiled optimized; compile some code optimized by default.
• Handle version numbers with additional digits.
• Avoid deadlock.
• Needless stop() call.
• Bug: callcc code crashes when compiled optimized.
• Efficiency: needless copies of field_matchers.
• Enable contains to more types of stackstrings.

 

[satesco]

I marvel at your ability to find solutions and improve your app. Probably, to exaggerate a bit, that chip Stid135 is a kind of "intellectual aphrodisiac" for your mind.
I've read the improvements and new features for the subsequent releases. Just reading it takes time. What about working as such to fix bugs or add something extra to your neumoDVB?
I can imagine that this project eats up your time, so I have the reluctance to suggest you consider for the future, for 6903x/6909x cards, the possibility to indicate the presence of generic streams, and if it is not an impossible claim, at least to display information of these streams, as you can see with Crazyscan under Windows, a mux with S2X extension to 10.0E.

 

[ptergrand]

When will the New Version be released?

 

[deeptho]

The next release is available. It has been tested on fedora 38 and on ubuntu 23.10.

One problem that is still not fixed is that it is not possible to view multiple services at once.
This is a bug in the mesa library. I reported it 6 months ago, but nobody seems interested
in fixing it.
The bug was reported here.

kwin patch causes program using two gl windows to crash (#9125) · Issues · Mesa / mesa · GitLab

System information System: Host: streacom.mynet...

gitlab.freedesktop.org

Tip: Perhaps if other users ask for a fix, eventually it might be fixed.

 

[ptergrand]

Yes, I have the same problem in Linux, the system crashes when I open more than one image.