Skip to contents

Changelog

Source: NEWS.md

orgdata 1.5.0

  • Make HDIR-systems default, keep FHI as an option.
  • Some internal functions made more general to be used in other packages

orgdata 1.4.9

  • geo_merge() gains a localtable argument. It can now be used on mapping tables generated with geo_map_multi.
  • Add geo_map_multi(), to generate a multiyear mapping table. 9b3e1a1
  • Fix geo_merge() d163c9a
  • Temporarily made it possible to run on both FHI and HDIR systems.
  • Allow specification of encoding in read_file ba980a7

orgdata 1.4.8

  • Check filegroup input is correct and give suggestion if not found db19c0d.
  • Fix debug.rows options.
  • Check global options are updated a10ced4

orgdata 1.4.7

  • Can use different encodings for reading files and Access database. Encoding can be specified in options with either orgdata.encoding.access or orgdata.encoding.csv. Check config file.
  • Fix geo splitting in geo_recode() using fix=TRUE as default argument.
  • Change options name from debug.row to debug.rows.
  • Fix problem updating global options when config file has changed (#320)

orgdata 1.4.6

  • Should not give unnecessary warning with dummy_grk (#318)
  • Rename argument year to year.geo in make_file() to be more explicit (#317)

orgdata 1.4.5

  • Have access to future data from API (#316).
  • Use pak package for upgrade.

orgdata 1.4.2

  • Display output when called.
  • Require norgeo package version 2.3.1

orgdata 1.4.1

  • Refactor some codes.
  • Specify encoding via config.yml file.
  • Use norgeo package version 2.3.0 from Github ie. dev, instead of CRAN.

orgdata 1.3.0

  • Fix selecting files without KOBLID (#315)
  • Add more testing.
  • Avoid warning when connection to the database is already closed.
  • Create environment for status flow ie. orgEnv.
  • Can use select arg to ease choosing file(s) in make_file() instead of KOBLID. Read the document on how to use the argument.
  • Refactor some codes.
  • Edit some documents.
  • Fix dbDisconnect error (#312)
  • Edit message (#313)
  • Edit few other messages as well.
  • Manage encoding warning for R 4.1 and below.
  • Filtering with IBRUKTIL and IBRUKFRA for selecting original files. IBRUKTIL uses >= and IBRUKFRA uses < of the specified date. (#309)
  • Update vignette on debugging.
  • Utilise input from config when mutating columns #ref
  • Add welcome logo.
  • Use integer (#310)
  • Proper use of boolean (#311)

orgdata 1.2.0

  • Update vignette.
  • Use keyword delete or slett when recoding with RE options to "". This replaces empty and tom as in #285.
  • Selecting geo levels when geo codes derived from two columns will use the maximum number of digits to find geo level (#307)
  • Rewrite some codes for speed. (#231)
  • Add unidentified municipalities as xx99 to geo_map(). Unidentified municipalities do not exist from API. (#304)
  • Age group with AgeCat can use mix categories easily with [x]. See example in find_age_category(). (#305)
  • Fix bugs when writing and reading data from warehouse ie. marked the KONTROLLERT column in Access (#298)
  • Fix bugs when accessing or deleting data with see_data() from data warehouse (#300)
  • Upgrade package version dependency.
  • Create unknown municipality codes with xx99 in geo_recode() function ie. geo record table in Access. (#302)
  • Recode unknown municipality codes with xx99 for known county. (#303)
  • MANHEADER accepts regular expression to select columname in addition to column index (#301)
  • Function read_file() accept SPSS file too (#297)
  • Function read_file() now accept txt file extension.

orgdata 1.1.0

  • Make geo abbreviation clearer in the global options and no more depending on the order of geo names (#286)

orgdata 1.0

  • Order does matter when using functions in EXTRA column (#294)

  • Fix S3 age category functions find_age_category() (#294)

  • Delete unwanted column in the output when using AgeCat function in EXTRA (#295)

  • Use global options in input argument when relevant instead of using match.arg() (#263)

  • Edit document here and there.

  • Use codebook to recode age category for better speed (#292)

  • Read or delete data in the data warehouse can use multiple koblid in see_data(). Use "all" in koblid argument to select all data on the chosen filegroup (#291)

  • Fixed duplicated age groups (#290).

  • Use either empty or tom to represent regular expression to replace to since Access makes symbol "" to be invisible (#285)

  • Change function name from see_org() to see_data() for viewing data in the data warehouse.

  • Use symbol | to separate multiple arguments in column EXTRA (#288)

  • Group age to specific or specified interval with AgeCat(). This function can be use in table for filegroup under EXTRA column (#287 #289)

  • Delete raw of similar columns with multiple specifications (#282)

  • Different ways to recode of similar column ie. duplicated, with defined lesid will give error.

  • Show current installed version against the new release version.

  • Some minor text editing.

  • Delete older file of DuckDB automatically since new version can’t read the older files (#280)

  • Actively ask users to update with given options when new release version is available.

  • Control all columns that should be numeric don’t contain any string. The columns to be controls for is now dynamic (#281)

  • Rename function alias from rf() to rdf() since rf() is already in use in stats package.

  • Deleting row with specification of lesid and without lesid ie. common filegroup, create errors. This is not fixed (#279)

  • Need to update version whenever a new release is available to ensure everyone is using the latest release. Else users can’t load the package.

  • Filter active files with date was done pragmatically ie. all files with BRUKTIL date other than 01-01-9999 will be excluded. Now filtering with date will be compared against current date. (#272)

  • Delete dataset from the database when unmark column KONTROLLERT instead of updating the dataset due to time consuming by updating it. The users have to mark the column to save or read the dataset in the database. (#278)

  • Change default for argument raw in make_file() to FALSE as in config file.

  • Some text editing here and there.

  • Function is_colour_txt() can specify symbol directly without needing to rely on the global options or to use withr package. Just for cosmetic purposes 😃

  • Standard columns should be uppercase. Using lowercase creates error when GEO comes from two separated columns especially when handling recode from codebook (#277)

  • Standardize arguments names in see_org() function.

  • Parallel processing is deactivated since it doesn’t work smoothly with Access and DuckDB connection ie. DBI package, due to the problem with not-exportable objects for future package.

  • Replace SQLite with DuckDB again due to the speed for DuckDB. But since DuckDB is under active development, the new version might not work with the file created by older version. In such situation older DuckDB database should be deleted and re-run with the newer DuckDB version.

  • When loading the package, users will be reminded whenever new version is available.

  • Show as data frame when warning NAs with coercion instead of just the GEO number where the coercion took place (#274)

  • Replace DuckDB with SQLite (#271)

  • Columnames are case insensitive (#115)

  • Use ellipsis for other arguments of data.table::fwrite() in save_file() function.

  • Implicit null for number of geo digits is moved to config file. This make it easy to maintain and expand.

  • Use future.apply package conditionally to reduce package vulnerability.

  • Deactivate dependency packages for parallel processing. Use it conditionally ie. to install manually if needed.

  • Reshape multiple columns containing more than one VAL is handled properly. When reshaping multiple columns to be just one VAL then leaving RESHAPE_VAL empty ie. use all columns not defined in RESHAPE_ID should work as before (#269)

  • Give more information when error to reshape data, especially long reshape. Most of the time the source of error is in defining RESHAPE_KOL and RESHAPE_VAL (#268)

  • Can reshape multiple columns when reshaping to LONG. Applicable when there are multiple columns represent number of cases and one column represent denominator for all these numerators (#266)

  • Use KOBLID to add or delete from Duck database (#261)

  • Depends for norgeo package refers directly to CRAN version instead of GitHub repo.

  • Use function geo_merge() to add geo granularity that aren’t from API. This can be a csv, xls or any other format that is accepted by read_file (#262)

  • Deactive KONTROLLERT without the need to umark from Access database by using argument raw = TRUE when using function make_file() (#264)

  • Use KOBLID as table name in DuckDB instead of FILID since FILID is not a unique number ie. can be used by multiple KOBLID.

  • Update text and website

  • Change function name from geo_levels() to geo_map() for mapping geo codes granularity.

  • Change helper function names for read_file(). (#250)

  • Fixed bugs when reading file from the web with https (#251)

  • read_file() accept Stata file with dta extension (#252)

  • Use yaml format for global configuration file to ease update. The file is located in config repo (#256)

  • Raw data that have been controlled for are saved in DuckDB database format. This increase reading speed especially for big files. Mark column KONTROLLERT in the Access registration database to activate this function. Unmark to read from the original raw file instead. (#257)

  • Use see_org() to read the raw data in the database. Argument action = "delete" can be used to delete the data from the raw database.

  • Function geo_merge() for merging geo codes that aren’t available from API to the mapping table ie. tblGeo, in the geo database. The data could be in any file format accepted by read_file() function. The data to be merged must have column to be merged ie. id.file, that is equivalent to the column id in the database ie. id.table. The id.file must be unique.

  • Handle unbalanced parentheses in post recode whenever possible when raw is used, else give error message. #246

  • read_file() accept filegroup name as argument in file to read the completed file after running make_file() function. #247

  • Debug functions can go deeper to show helper functions as well with debug_opt("deep") or options(orgdata.debug = "deep") #243

  • Post recode uses type PS in codebook. The function is used when there is a need to recode the value of a column after the dataset have been clean and aggregated. Specification to select the row to be recoded uses either standard expression or R syntax of data.table style. When using R syntax the value must have raw prefix eg. raw(AAR %in% c(2000, 2005)). #244 #245

  • Delete rows with "-" minus symbol in TIL column in the codebook is accepted for do_recode_post().

  • Able to aggregate to country level in addition to other geographical levels (#240)

  • Download the geo code to aggregate with only once despite the different geographical levels the dataset to be aggregated into (#241)

  • Update package via function update_orgdata(). Basically it’s just a wrapper for remotes::install_github().

  • Can aggregate to country level as well (#240)

  • Speed up aggregate process (#241)

  • Fixed the bugs with reshape wide consisting multiple dimensions ie. more than one TAB columns. (#228)

  • Ensure melting columns are not converted to factor (#234)

  • Some columns should be numeric type and these columns are GEO, AAR, KJONN, ALDER and all VALs. They are also specify in global options orgdata.num. Ensure thise columns are numeric and give warning as well as log when coercion where NA is introduced. (#235)

  • Use log file to check what or which codes that have problem. (#237)

  • Warn and give logfile if GEO can’t be converted to integer before geo recode. (#233 #236)

  • Able to read files that have no specific extension but they have to be coma or semicolon separated (#227)

  • Ensure all VAL columns are numeric type (#229)

  • Give explicit warning when GEO codes have character that can cause coercion resulting in converting GEO codes to NA (#229)

  • Log files for code99 also include koblid. The files will be named as code99_koblidxxx (#222)

  • Running parallel processing is unstable. Not sure if the process crash due to MS Access that can’t handle constant requests or … Anyway, running parallel now will only use 50% of the available cores (#224)

  • User can specify more or less cores than 50% if needed. To use 75% cores by specifying parallel = 0.75 in the argment make_file() or in the global options orgdata.parallel (#225)

  • Connect to both database with common function. (#212)

  • Use global options orgdata.year to specify production year if not using current year. (#216)

  • Use parallel processing in make_file() with argument parallel = TRUE. (#217)

  • Get feedback for successful and unsuccessful filegroups when running make_filegroups(). (#199)

  • Reshape wide is not limited to only 3 reshape columns as it was. (#200 #201)

  • Create log files in default orgdata folder instead of occupying the REM. (#202)

  • Delete old bydel codes ie. before 2003, except for Oslo. This is due to the unstable geographical recoding of bydel. Use EXTRA column on filegroup level with argument DeleteOldBydel. (#204 #206)

  • Log files use KOBLID to be more specific. (#208)

  • Reshape to wide format can now implement multiple reshape dimensions which are usually one or multiple TAB columns. For instance for LESEFERD of tot_elev is the product of both mestringsnivå and klassetrinn. (#188)

  • The output will be arrange on a standard column order. (#190)

  • Debugging will be easier with a wrapper function debug_opt(). (#196)

  • Fixed #147 to implement reshape wide format to create denominator. This function is only applicable to certain file structure received from SSB. For example filegroup LESEFERD where column tot_elev represents the total number of student with mestringsnivå and not the grand total of students. The number of students with mestringsnivå is represented in column ant_elev. Therefore the long format for mestringsnivå needs to be restructured to wide with value from the ant_elev to ensure summing up tot_elev when creating denominator will not create a grand total of students instead of the total number of students with mestringsnivå. (#184)

  • Fixed #162 when saving geographical codebook with write = TRUE and the table doesn’t exist in the geo-code database.

  • Stop looping the geographical levels when orgdata.debug.geo or orgdata.debug.aggregate are active and make the default to kommune. (#166)

  • Add batch date from codebook as reference when debugging. (#168)

  • Hide warning messages if file has been controlled for any possible errors. It means the column KONTROLLERT in table for original files is marked for the specific file. (#171)

  • Refactoring function for recode of geographical codes. (#174)

  • Geo codes for enumeration areas and towns that aren’t able to be merged will be converted to unknown codes with either xxxx9999 or xxxx99. (#177)

  • Create default folder at C:/Users/YourUserName/orgdata_logs when path argument is not specified in save_file(). (#179)

  • Create unknown codes for enumeration areas and town codes if not able to recode. Most probably is due to the enumeration codes before 2002. The Excel file from SSB is not clean and unstable to be used for recoding. The unknown codes are based on municipality codes with added xxxx99 or xxxx9999. As in #177 but recode is done on municipality codes before merging back to the original dataset. (#182)

  • Add codeDelete in log for geographical codes that aren’t able to be merged. The codes will be excluded in the dataset. To access all the deleted codes use log$codeDelete. (#149)

  • Give explicit error message when path is missing in save_file(). (#152)

  • Able to select any valid year to recode geographical codes and aggregate accordingly. But recoding geographical codes backward isn’t possible. (#153)

  • Error message will split a long vector if exist. (#153)

  • Add column with batch date when the geographical codes were downloaded form SSB when creating a geo codebook with geo_map() or geo_recode(). (#156)

  • Actively select only the first geo code when old geo code is split into multiple new geo codes. (#159)

  • Able to select a base year for geo recode based on the year available in the original data. This is available in column AAR in the dataset. Use argument

    base or global option orgdata.recode.base with logical input. TRUE will select the base year for recoding geographical code from the year of the original file to the current year. Default is FALSE ie. include all available geographical codes available in the codebook. (#157)

  • Fixed #139 for orgdata.debug.geo keep original geo codes for enumeration areas before adding 9999. (#140)

  • Fixed #142 show codes that have problem to recode directly instead of row numbers (#144)

  • Save all codes that have problem in log environment for easy access. To list the codes is either with log$code00 or log$codeShort

  • Recode geo even when argument aggregate = FALSE in make_file() function.

  • Rename make_filegroup and lag_filgruppe to plural ie. make_filegroups and lag_filgrupper.

  • Use options orgdata.debug.rows to select only specific row(s) for further processing. It can be activated via global options with options(orgdata.debug.rows = 20:50) or via argument row = 20:50 in make_file() to select row 20 to 50.

  • Fixed #135 with incorrect geo recode. (#131)

  • Make multiple filegroups via make_filegroups. (#137)

  • Fixed #132 LANDSSB must be string

  • Convert whitespace to NA to be able to delete all rows with NA

  • Fixed #119 able to mutate for TABS and VALS as well (#126)

  • Fixed #122 delete rows with NA via EXTRA column (#127)

  • Fixed #118 warning text when column(s) aren’t defined in FILGRUPPE and will be deleted (#128)

  • Edit error message for columns with existing NA value before aggregating. Total value will be NA and this will conflict with the allready existing NA category in the aggregated column(s). Therefore existing NA value in the selected column(s) must be recoded to a valid value.

  • Use options(orgdata.debug.geo = TRUE) to keep old geo codes for debuging (#120)

  • Use reset_options() to reset to default options.

  • Warn when process discontinued due to debugging.

  • Add vignettes for Standardize git and Debugging

  • Fixed #121 recode geographical code for split codes (#120)

  • Change database filename to raw-database_BE.accdb

  • Fixed #106 split long messages (#107)

  • Fixed #108 #112 grunnkrets codes that have changed before 2002 not available via API from SSB while code changes for municipality includes changes from 1977. Check from SSB website. We use the municipality codes to create uspesified grunnkrets codes for data before 2002 (#109 #113)

  • Fixed #110 updating SQL code for new table name for codebook (#111)

  • Check columns to aggregate for any possible NA (#98). Columns that have NA should be recoded to uoppgitt or something equivalent since leaving the category to NA will conflict with NA representing total value when aggregating.

  • Fixed #100 for grunnkrets that ends with 00 have no correspond codes from SSB API. Need to add it manually (#101)

  • Fixed #99 when geo codes fails to be recoded then the row index will be shown (#103)

  • Geo codes ends with 4 zeros xxxx0000 neither have equivalent codes from SSB nor representing a correct coding structure as so called Delområde that ends with 2 zeros xxxxxx00. To avoid missing the information, these geo codes are recoded to xxxx9999 with function is_grunnkrets_0000() as in PR (#103).

  • see_file() accepts just a single numeric as well.

  • Fixed #85 see_file() list all the columns when columnames or column indexes are not specified. The variables are sorted whenever possible. (#87)

  • Add more function tests (#88)

  • Exclude LANDSSB in aggregate when split to LANDBAK and INNVKAT. This is because code 0 will be recoded to 20 when split and causes unnecessary more rows (#84)

  • Delete deprecated functions.

  • Fixed #93 when source level can’t be identified due to NA.

  • Fixed #95 for grunnkrets codes that aren’t missing but have less number of digits ie. less than 7 digits. Assuming these are codes for municipality then 9999 is added at the end of these codes (#96)

  • Gives row number for GEO codes that get coerced as NA when converted to integer. This will make it easy to check in the original raw data (#96)

  • Aggregate now give total to all dimensions including those specified in AGGKOL (#82)

  • Function see_file() accept column index as well (#83)

  • Recode variables using regular expression when defined in codebook with type RE. Finding pattern can either be written in ordinary regular expression ie. \\d{4}.* or with rex() package. (#78)

  • New feature for checking categories for variables with see_file() (#75)

  • Fixed #65 make TABS and VALS dynamic for easy extension for these columns (#66)

  • Fixed #64 recode of variable that has different class (#68)

  • Fixed #63 implicit null includes all possible VAL columns when exist (#69)

  • Fixed #70 recode GEO of different object class (#71)

  • Fixed #67 aggregate with total values for standard variables ie. UTDANN, LANDSSB, LANDBAK and INNVKAT (#72)

  • Fixed #61 use AGGKOL in Access registration database to specify other columns to aggregate other than the standard eg. KJONN, TAB1, TAB2 etc. (#73)

  • Fixed #55 to recode standard variables via codebook instead of hard coded (#58)

  • Fixed #52 skip split if not specified (#59)

  • Fixed #57 split column with duplicated values will keep the original column (#60)

  • Fixed #56 aggregate all VAL columns whenever specified and not only specific to VAL1 (#62)

  • Edit verbose messages

  • Reshape dataset from wide to long. Reshape can have more than one measure variables. Please read how this is specified in Access registration database.

  • Split columns must have equal number of values to the defined SPLITTIL. Duplicate the value if it is less than the maximum SPLITTIL. For example for value 0 in column LANDSSB which will be split into LANDBAK and INNVKAT, the value will be duplicated into 00 to avoid split with value NA.

  • Recode for LANDBAK and INNVKAT after aggregating are done internally ie. hard coded, in do_aggregate_recode_standard(). Total is coded with 20. Any eventuality for future change should also look other related functions such as is_aggregate_standard_cols() and is_col_num().

  • Change argument parameter for find_spec() function.

  • Update text document in several places.

  • Add colour type warn2 for warning message without Warning: prefix.

  • Request (#43) messages with specific colour

  • Fixed (#46) recode to string even though columns is type integer or numeric.

  • Unknown bydel ie. (uoppgitt) is added when enumeration areas codes ie. (grunnkrets) for bydel is XXXX9999 in function geo_level().

  • Add unknown grunnkrets for kommune when not available since some of the datasets have unknown grunnkrets that aren’t listed in API downloaded data (#39).

  • Exclude TAB1, TAB2 and TAB3 from being aggregated. (#44)

  • Recode for aggregated variables uses AG in TYPE column in the codebook instead of FILGRUPPE with AGGREGATE as it was implemented in ver 0.2.0. This will make it possible so specify FILGRUPPE and LESID to implement the principle for GENERAL, COMMON and SPECIFIC variables.

  • Change function name do_aggregate_recode to do_aggregate_recode_standard for standard variables.

  • Recode for aggregated categories can be defined in Recode form ie. codebook, and use AGGREGATE in the specification under FILGRUPPE

  • Delete rows when defined in codebook using minus symbol under TIL column. Similar principles is implemented for GENERAL, COMMON and SPECIFIC feature as in recode. Read detail in ver 0.0.5 - alpha.

  • Display both columnames to be recoded that are found in the dataset or those that aren’t found when defined as ALLE in the codebook so user will be aware of its existence.

  • Standardize some most used arguments to read_file() such as nrows, header, skip, trimws and na. Read details in read_file() function description.

  • Output to read_file() as data.table class.

  • Use standard columnames with V1, V2 etc when argment header = FALSE is specified.

  • Error message with list of unmatch columns in do_column_standard().

  • Give clearer message and debug message eg. Execute: read_file().

  • Change MAPPE to UTMAPPE to make it more explicit for path specification to save file.

  • Defun orgdata.active global options to use columnames from original dataset.

  • Use global options options(orgdata.debug.nrow = TRUE) to read only first 20 rows. Suitable for debug purposes.

  • Fix (#28) GEO derived from two columns with empty INNLESARG.

  • Add column LEVEL for granularity level ie. grunnkrets, fylke, kommune, bydel etc

orgdata 0.0.1 - alpha version

  • When MANHEADER is used then the new columname must be specified in the respective standard column (#21)

  • Deprecated arguments geo and val in make_file. Output data must use standard columnames instead of keeping the columnames from original dataset.

  • Rename functions read_raw or lesraw to make_file or lag_fil (#27)

  • Alle functions uses underscore “_” for both english and norwegian.

  • Add new columns if one of the standard columns is missing in the original data. The value to be inserted to the new column must use symbol less than < and more than >. For instance when column KJONN doesn’t exist in the original data, we can specify with <2> in under column KJONN in the Access registration database. The output will add a new column KJONN with value 2. (#15)

  • Fix #13 and #18

  • Default orgdata.verbose is TRUE.

  • Options for orgdata.implicit.null with default as TRUE. Use options(orgdata.implicit.null = FALSE) to deactivate (#19)

  • Changes is in PR #11 and #12 is as the following:

  • Implicit zero (#11). Discussion is in Gist

  • Use version specific for imported packages.

  • Rename standard column LANDBAK to LANDSSB for column in original data received from SOB containing information about country of origin.

  • Save file as specified in column MAPPE in Access registration database or specify in path argument for function save_file. (#12)

  • Changes is in PR #8 is as the following:

  • Create GEO code from two separate columns. This has to be defined in Access registration under GEO with comma separated eg. nameGeoCol1, nameGeoCol2.

  • Order standard columns in the output dataset with this order for the first four columns:

    • GEO, AAR, ALDER, KJONN

  • Change norwegian name for save_file from lagfil to lagrefil.

  • Use column name KOLNAVN instead of ADDKOL.

  • Rename function do_addcols and get_addcols to do_colname and get_colname to be consistent with the changes in Access registration database.

  • Changes is in PR #4 is as the following:

  • Recode variables from specification in tbl_KodeBok uses:

    1. GENERAL variables are defined in FILGRUPPE as ALLE and are used to recode variables in all groups.
    2. COMMON variables are when FILGRUPPE is specified but have empty LESID. This will recode variables within selected group.
    3. SPECIFIC variables are when FILGRUPPE and LESID are specified. This will recode variables in that specified FILGRUPPE of the specified FILID.

  • When all these three specification exist in tbl_KodeBok:

    • SPECIFIC variables will overrule COMMON variables
    • COMMON variables will overrule GENERAL variables

  • Write as <NA> in codebook under column FRA when specifying missing variables indicating that a missing column to be recoded to value in column TIL. This will differentiate between real missing and a real column value of NA. (#5)

  • Error message will be given if LESID is specified without FILGRUPPE since LESID is not unique ID.

orgdata 0.0.4 - alpha

  • MANHEADER and KOLNAVN uses common helper function is_col_separate()
  • MANHEADER old convert to integer and use and index for columns

Changes is in PR #2

orgdata 0.0.3 - alpha

  • LESID is not unique id but a combination of LESID and FILGRUPPE
  • Standard columns include 3 TABS and 3 VALS
  • KOLNAVN input uses comma as separate and VAL1=TOTAL, TAB1=ICDis valid input

Changes is in PR #1

orgdata 0.0.2 - alpha

Things that are implemented

  • Bla.. bla.. bla..