The app is a smaller download, launches faster after the first run, and holds up better during a long session. This release also picks up a month of analysis-engine work, and that part is not routine. Results, messages and printed output move closer to SAS. Some programs that ran clean in 1.0.0 now stop with an error, and some numbers change. Please re-run anything you rely on before trusting its output. THE APP - Jenner is hundreds of megabytes smaller to download and to keep on disk. - After the first launch, Jenner opens noticeably faster. The first launch after installing or updating takes slightly longer, once. - Exporting several notebooks to PDF at once is faster. - The file list settles sooner at startup and redraws less often while you work. - If Jenner's workspace shuts down unexpectedly and leaves a blank window, Jenner reloads it instead of leaving you to quit and start again. Notebooks come back from their last save, and Jenner tells you when that happens. - Export to PDF, Reveal in Finder and subscription checks no longer look stuck after a reload. PROGRAMS THAT RAN BEFORE MAY NOW STOP - A DATA step naming a format that cannot be resolved now stops with an error, as SAS does. NOFMTERR restores the old permissive behavior. - Combining data sets that disagree on whether a variable is text or numeric now stops the step instead of silently writing missing values. - PROC APPEND stops on a character-versus-numeric mismatch, or a longer incoming character variable, unless you use FORCE. - PROC SQL rejects BINARY and VARBINARY column types instead of quietly discarding the values, and refuses a function call written with no arguments. - Database, web-service and AI connections must now point outside your own machine and local network. A model server or database running locally worked before and no longer does. File-based and in-memory databases are unaffected. - Some ways of reaching files outside your workspace folder are now refused. - A clustering or distance analysis too large to build stops with an error instead of exhausting memory. Splitting the data with BY completes the work. RESULTS AND PRINTED OUTPUT - Classification variables now default to the last level as reference in several modeling procedures, and logistic models default to effect coding. Estimates for such models will differ from the previous release. - Corrected results where the computed value did not match SAS: listwise deletion in correlations; Type III sums of squares and multiple-comparison limits in general linear models; the Mantel-Haenszel chi-square and the phi coefficient in frequency tables; weighted percentiles, variance, skewness and kurtosis; weighted standard error, coefficient of variation and sums of squares; character functions in SQL. HMAC-SHA256 took its key and message in the wrong order and now produces the standard digest. - A period in a text column now counts as an ordinary value in frequency tables, so counts and percentages shift for data that uses it as a code. - Padded text compared against a number now matches, so some programs keep rows they used to drop. - Numeric output follows the SAS width rules, LEFT and PUT pad as SAS does, PROC PRINT numbers observations from the source data set, and the PROC CONTENTS output data set carries the columns SAS produces. - Table layout, headings and charts follow SAS more closely, and macro statements run in SAS's order across nested macros. - Reading data containing accented letters and other non-ASCII characters no longer fails the DATA step. REMOVED - PROC DS2 has been removed. It did not run in earlier versions either: a DS2 step reported the procedure unavailable and produced no data, and a program that called package methods from a later step finished but wrote values that were not correct. Those programs now stop with an error instead. Most DS2 programs map directly to a DATA step, with package methods written as PROC FCMP functions.