Changes, bugs, and tasks

– version 2.1.0 / CGI 1.1.0: cgi write access restored
* other minor improvements:
– Unidatab.css etc.
– version 2.0.x: a major update involving structural changes.
* table node split in two tables: unid_node and unid_symbol
old dbs can be updated to the new standard (see doc/migration.txt).
* unid-adm now works on 2 local connections at a time.
In the future this feature shall allow copies of records
between databases, and remote connections will be implemented.
* table renaming : for compatibility with other dbs
* timestamps with milliseconds
* several new SQL functions added:
– ana_base() — recursive analysis of a base and graphic output
– ana_record() — recursive analysis of a record
– duration() — check the time spent on one query
– find_format() — better extraction of a record’s format
– size() — get the size of some data or record
– time_flip() — convert from and to time stamps
– uid() — get the id of some user (by name)
– uname() — get the name of some user (by id)
* simplification and improvement in unid-adm command line syntax:
– lastdata targets various containers and can write to file
– change 1 (or 2): to connect dbs
– use 1 (or 2): to change db in use
* replacement of OpenSSL by CryptoPP
171017: Unidatab-CGI minor changes
—  minor fix in Params_Sql
—  more error messages
status: done
171017: Unidatab-CGI : multi-user failure
— a user being logged in for some database, the same database is served for another user
status: fixed

The CGISESSID cookie being the same for Unidatab.cgi only one user for one database was allowed. This has been changed thru better cookie specification.

171016: Unidatab-CGI : CGI failure
— when a radio button is not checked as it should
status: fixed

Some actions – like symbol:modify – require the node selection radio button to be checked. Otherwise the request sent yields a CGI failure.

171016: Unidatab-CGI first bugs with utf8 and wide chars
—  encode-decode problems
status: fixed

Always testing the cgi with English strings made me miss UTF8 weaknesses in the code. I only noticed these bugs installing Unidatab on my mother’s machine!

171015: Unidatab-CGI default cgi directory
—  set to /cgi-bin/ and added a new $UNID_CGI_ROOT global variable for convenience
status: fixed
171012: Unidatab-CGI version 1.0.x
— independant repository created

Cf. and the related blog post here.

170914: version 1.0.0 released

(see the related blog post)

170806: ComPP-1.7 doesn’t build under glibc 2.25
— unexpected segmentation fault with TCP and UDP servers
status: not fixed

However one can still use the libraries built under an older version of glibc in an environment using glibc 2.25 (like Fedora 26) .

170212: RecordModify() bug in unid-gui
— the function didn’t preserve subdb format
status: fixed

UNID_CODE_SUBDB was suppressed by filterQuotes(). Note that the best way to replace a format by another in unid-gui is to provide a valid (positive) format id, be it done within the input mask of menu ‘Modify’ or more directly via Reformat.

170112: gtk3 related bugs in unid-gui
— CPU spends too much activity in the beggining of execution
status: not done

CPU overload stops as soon as a row is copied (for instance), unfortunately the problem starts again as soon as a record is displayed. See the BUGS page on Github for stderr output. See this question with more details on WxWidgets Discussion Forum. In consequence, we choose to downgrade wxWidgets to 2.8 for our build until  a solution is found (see INSTALL).

170112: some bugs fixed in unid-gui
— copying rows to lists
— better menu layout
status: done
 170105: An old concern
— an attempt to confining administrative tasks to unid-adm dates back to 2013
— never found correct implementation  (see : sqlite_do())
status: not done

Traces of such concerns (legitimate or not) may remain as unused code specs or murky comments.

 170103: Makefile and new directory tree
— old compilation recipes still correct
status: done
 161129: revision of get_subject(continued)
— set_myvar included: no preamble needed anymore
status: done
– details in last_out and unid-adm
– details in get_subject()
– multi-line input (maybe introduced a bit earlier)
status: done
161114: version 0.4.3 : revisions and unidSql improvements
— search functions
— help strings
— introduction of last_out()
status: done
 161105: version 0.4.3 released:
status: partly tested
 160827: UnidSql: bug: swap_links():
severity: high: corrupted db
status: corrected (160830)

swap_symbols() did the job anyway.

160827: UnidSql: bug: swap_symbols():
severity: medium
status: corrected (160830)

User and time information got corrupted for swap_symbols() didn’t update nor swap them.

160821: Unidatab.cgi: bug: integer strings:
severity: medium
status: corrected (161009)

“(de)stringification” of integers, which is central in Unidatab, was unfortunately not always handled correctly.

160821: unid-gui: bug: DateTime format field:
severity: low
status: corrected (161105)

didn’t toggle.

160820: unid-adm: unfinished function: replace_symbol() :
severity: low
status: not done

some modes are not coded. At present replace_symbol() only accepts mode=0 (changes symbol in one specified link location within some record), or mode=-1 (changes the symbol everywhere).

160718: unid-adm: bug: sql_help(‘make_full_record’):
severity: low
status: corrected (161105)

returned a wrong definition (3 params instead of 4); correct definition:

160718: unid-gui: bug: editing a symbol:
severity: medium
status: corrected (161105)

in the ‘Edit’ tab, editing a symbol didn’t consider symbols with record references.