{"id":121901,"date":"2025-02-13T14:22:07","date_gmt":"2025-02-13T12:22:07","guid":{"rendered":"https:\/\/prohoster.info\/blog\/novosti-interneta\/reliz-subd-duckdb-1-2-0"},"modified":"2025-02-13T14:22:07","modified_gmt":"2025-02-13T12:22:07","slug":"reliz-subd-duckdb-1-2-0","status":"publish","type":"post","link":"https:\/\/prohoster.info\/sq\/blog\/news\/reliz-subd-duckdb-1-2-0","title":{"rendered":"Publikimi i DBMS DuckDB 1.2.0","gt_translate_keys":[{"key":"rendered","format":"text"}]},"content":{"rendered":"
\u00cbsht\u00eb publikuar versioni 1.2.0 i SGBD-s\u00eb DuckDB, e orientuar p\u00ebr ekzekutimin e pyetjeve analitike dhe konceptualisht e ngjashme me SQLite. DuckDB kombinon ve\u00e7orit\u00eb e SQLite, si kompakt\u00ebsia, lidhja n\u00eb form\u00ebn e nj\u00eb biblioteke t\u00eb integrueshme, ruajtja e DB-s\u00eb n\u00eb nj\u00eb skedar t\u00eb vet\u00ebm dhe nd\u00ebrfaqja CLI, me mund\u00ebsi dhe optimizime p\u00ebr ekzekutimin e pyetjeve analitike q\u00eb p\u00ebrfshijn\u00eb nj\u00eb pjes\u00eb t\u00eb konsiderueshme t\u00eb t\u00eb dh\u00ebnave t\u00eb ruajtura, p\u00ebr shembull agregimin e gjith\u00eb p\u00ebrmbajtjes s\u00eb tabelave ose bashkimin e disa tabelave t\u00eb m\u00ebdha. Kodi i projektit \u00ebsht\u00eb shkruar n\u00eb C++ dhe shp\u00ebrndahet n\u00ebn licenc\u00ebn MIT. <\/p>\n
DuckDB ofron nj\u00eb dialekt t\u00eb zgjeruar t\u00eb SQL, q\u00eb p\u00ebrfshin mund\u00ebsi shtes\u00eb p\u00ebr p\u00ebrpunimin e pyetjeve shum\u00eb komplekse dhe me koh\u00eb t\u00eb gjat\u00eb ekzekutimi. Mb\u00ebshtoret p\u00ebrdorimi i tipeve komplekse (vargje, struktura, bashkime), si edhe ekzekutimi i n\u00ebnpyetjeve arbitrare dhe t\u00eb nd\u00ebrlidhura t\u00eb futura brenda nj\u00ebra-tjetr\u00ebs. Mb\u00ebshtoret ekzekutimi i nj\u00ebkohsh\u00ebm i disa pyetjeve, ekzekutimi i pyetjeve drejtp\u00ebrdrejt nga skedar\u00eb n\u00eb formatet CSV dhe Parquet. \u00cbsht\u00eb gjithashtu e disponueshme mb\u00ebshtetja p\u00ebr import nga SGBD PostgreSQL. <\/p>\n
Projekti p\u00ebrdor nj\u00eb mb\u00ebshtjell\u00ebs nga SQLite, parserin nga PostgreSQL, komponentin Date Math nga MonetDB, realizimin e tij t\u00eb funksioneve dritare (t\u00eb bazuar n\u00eb algoritmin Segment Tree Aggregation), nj\u00eb procesor p\u00ebr shprehje t\u00eb rregullta t\u00eb bazuar n\u00eb bibliotek\u00ebn RE2, optimizuesin e tij t\u00eb k\u00ebrkesave, mekanizmin MVCC p\u00ebr menaxhimin e ekzekutimit t\u00eb p\u00ebrbashk\u00ebt t\u00eb detyrave (Multi-Version Concurrency Control), si dhe nj\u00eb motor t\u00eb vectorizuar p\u00ebr ekzekutimin e k\u00ebrkesave t\u00eb bazuar n\u00eb algoritmin Hyper-Pipelining Query Execution, i cili lejon p\u00ebrpunimin n\u00eb nj\u00eb operacion t\u00eb grupeve t\u00eb m\u00ebdha t\u00eb vlerave nj\u00ebher\u00ebsh. <\/p>\n
N\u00eb versionin e ri: <\/p>\n
\n
E shpallur mb\u00ebshtetje p\u00ebr metodat e reja t\u00eb kompresimit, t\u00eb cilat nuk jan\u00eb p\u00ebrfshir\u00eb si parazgjedhje p\u00ebr t\u00eb ruajtur kompatibilitetin e skedar\u00ebve t\u00eb DB me versionet e kaluara t\u00eb DuckDB. P\u00ebr t\u00eb p\u00ebrdorur formatin e p\u00ebrmir\u00ebsuar t\u00eb skedar\u00ebve t\u00eb DB, \u00ebsht\u00eb propozuar mund\u00ebsia e lidhjes s\u00eb DB-s\u00eb me numrin e versions \u2014 kur hapni skedarin me parametrin \u00abSTORAGE_VERSION\u00bb, tani mund t\u00eb caktoni versionin minimal t\u00eb mb\u00ebshtetur t\u00eb DB-s\u00eb (\u00abATTACH \u2018file.db\u2019 (STORAGE_VERSION \u2018v1.2.0\u2019);\u00bb). P\u00ebr t\u00eb konvertuar formatin e ri n\u00eb t\u00eb vjetra, mund t\u00eb p\u00ebrdorni komand\u00ebn SQL COPY, p\u00ebr shembull: ATTACH \u2018file1.db\u2019; ATTACH \u2018converted_file.db\u2019 (STORAGE_VERSION \u2018v1.0.0\u2019); COPY FROM DATABASE file1 TO converted_file;\n
\u00cbsht\u00eb hequr kufizimi q\u00eb nuk lejonte shtimin e identifikuesve, p\u00ebr t\u00eb cil\u00ebt kishte nj\u00eb indeks p\u00ebr ndjekjen e unik\u00ebsis\u00eb, n\u00ebse k\u00ebta identifikues ishin fshir\u00eb n\u00eb transaksionin aktual. P\u00ebr shembull, kodi SQL i m\u00ebposht\u00ebm tani nuk \u00e7on n\u00eb gabim: CREATE TABLE students (id INTEGER PRIMARY KEY, name VARCHAR); INSERT INTO students VALUES (1, \u2018John Doe\u2019); BEGIN; DELETE FROM students WHERE id = 1; INSERT INTO students VALUES (1, \u2018Jane Doe\u2019);\n
\u00cbsht\u00eb shtuar mb\u00ebshtetje p\u00ebr ngarkimin e skedar\u00ebve CSV n\u00eb kodet Latin-1 dhe UTF-16 (m\u00eb par\u00eb mb\u00ebshtetej vet\u00ebm kodimi UTF-8). FROM read_csv(\u2018cities-latin-1.csv\u2019, encoding = \u2018latin-1\u2019);\n
\u00cbsht\u00eb shtuar mb\u00ebshtetje p\u00ebr p\u00ebrdorimin e ndar\u00ebsve me shum\u00eb byte n\u00eb skedar\u00ebt CSV (deri n\u00eb 4 byte), q\u00eb lejon p\u00ebrcaktimin e emoji si ndar\u00ebs kolonash. a\ud83e\udd94b hello\ud83e\udd94world FROM read_csv(\u2018example.dsv\u2019, sep = \u2018\ud83e\udd94\u2019);\n
Si parazgjedhje \u00ebsht\u00eb aktivizuar moda e rrept\u00eb e analiz\u00ebs s\u00eb skedar\u00ebve CSV (\u00abstrict_mode = true\u00bb), e cila kontrollon p\u00ebrputhshm\u00ebrin\u00eb e formatizimit me specifikimin RFC 4180. N\u00eb mod\u00ebn e rrept\u00eb, lejohet p\u00ebrdorimi vet\u00ebm i nj\u00eb simboli p\u00ebr ndarjen e rreshtave, q\u00eb do t\u00eb \u00e7oj\u00eb n\u00eb gabim gjat\u00eb p\u00ebrpjekjes p\u00ebr t\u00eb analizuar skedar\u00eb, ku p\u00ebr ndarjen e rreshtave p\u00ebrdoren jo vet\u00ebm simboli i rikthimit t\u00eb rreshtit, por edhe simboli i rikthimit t\u00eb karroc\u00ebs (\u00ab\r\n\u00bb).\n
N\u00eb parser-in CSV \u00ebsht\u00eb p\u00ebrdorur nj\u00eb algorit\u00ebm i ri p\u00ebr zbulimin e rreshtit t\u00eb ri, i cili mb\u00ebshtet paralelizimin e operacioneve. P\u00ebrdorimi i algoritmit t\u00eb ri e ka p\u00ebrshpejtuar analizimin e CSV me rreth 15%.\n
\u00cbsht\u00eb hequr kufizimi p\u00ebr madh\u00ebsin\u00eb e rreshtit n\u00eb skedar\u00ebt CSV (m\u00eb par\u00eb rreshti nuk duhej t\u00eb kalonte 8 MB).\n
Gjat\u00eb eksportimit t\u00eb t\u00eb dh\u00ebnave n\u00eb formatin Parquet \u00ebsht\u00eb shtuar mb\u00ebshtetja p\u00ebr hash-e (fjalor\u00eb) dhe indekse t\u00eb bazuara n\u00eb filtrin Bloom. \u00cbsht\u00eb zbatuar metoda e kompresimit DELTA_BINARY_PACKED, e cila mund\u00ebson ulje t\u00eb ndjeshme t\u00eb madh\u00ebsis\u00eb s\u00eb skedar\u00ebve Parquet.\n
N\u00eb nd\u00ebrfaqen e komand\u00ebs, \u00ebsht\u00eb shtuar nj\u00eb mod i sigurt, i aktivizuar me opcioni \u00ab-safe\u00bb ose komand\u00ebn \u00ab.safe_mode\u00bb. N\u00eb k\u00ebt\u00eb mod, lejohet qasja vet\u00ebm n\u00eb skedarin e DB-s\u00eb fillestar t\u00eb p\u00ebrcaktuar, dhe p\u00ebrpjekja p\u00ebr t\u00eb hapur skedar\u00eb t\u00eb tjer\u00eb do t\u00eb \u00e7oj\u00eb n\u00eb nj\u00eb gabim.\n
N\u00eb nd\u00ebrfaqen e linj\u00ebs s\u00eb komand\u00ebs \u00ebsht\u00eb p\u00ebrmir\u00ebsuar plot\u00ebsimi automatik i hyrjes. Kodi p\u00ebr plot\u00ebsimin automatik \u00ebsht\u00eb kaluar n\u00eb p\u00ebrdorimin e PEG (Parsing Expression Grammar).\n
Kur gjat\u00eb ekzekutimit t\u00eb komandave n\u00eb nd\u00ebrfaqen e komand\u00ebs, \u00ebsht\u00eb b\u00ebr\u00eb e qart\u00eb paraqitja e numrave t\u00eb m\u00ebdhenj, p\u00ebr shembull, kur shfaqet numri 100000000, do t\u00eb b\u00ebhet nj\u00eb shtes\u00eb e cila do t\u00eb thot\u00eb «(100 milion\u00eb)».\n
N\u00eb SQL \u00ebsht\u00eb shtuar mb\u00ebshtetje p\u00ebr sintaks\u00ebn, ku emrat e shkurtuar t\u00eb tabelave dhe shprehjeve mund t\u00eb p\u00ebrcaktohen p\u00ebrpara vlerave q\u00eb i referohen (n\u00eb vend t\u00eb p\u00ebrdorimit t\u00eb sintaks\u00ebs « shprehje AS emri »): SELECT e1: shprehje_disa_t\u00eb_gjat\u00eb_dhe_t\u00eb_nd\u00ebrlikuara, e2: t2.emri_i_kolones FROM t1: schema_e_gjat\u00eb.tabela_disa_t\u00eb_gjat\u00eb, t2: shkurt.tabela;\n
N\u00eb komand\u00ebn «SELECT» \u00ebsht\u00eb realizuar mb\u00ebshtetje p\u00ebr operacionin «RENAME» p\u00ebr rinovimin e fushave q\u00eb shfaqen kur p\u00ebrdoret shprehja «*»: SELECT * RENAME (col1 AS new_col1) FROM integers;\n
N\u00eb komand\u00ebn «SELECT» \u00ebsht\u00eb lejuar p\u00ebrdorimi i operacioneve «LIKE» dhe «SIMILAR TO» gjat\u00eb shfaqjes p\u00ebrmes «*»: SELECT * LIKE ‘val%’ FROM key_val;\n
\u00cbsht\u00eb p\u00ebrmir\u00ebsuar cil\u00ebsia e gjenerimit t\u00eb numrave pseudorast\u00ebsor\u00eb.\n
Optimizuesi i k\u00ebrkesave \u00ebsht\u00eb modernizuar. Performanca n\u00eb testin TPC-H SF100 \u00ebsht\u00eb rritur me 13%.\n
\u00cbsht\u00eb prezantuar nj\u00eb API i ri i ngjash\u00ebm me C p\u00ebr zhvillimin e shtojcave, i cili mund t\u00eb p\u00ebrdoret, p\u00ebr shembull, p\u00ebr krijimin e funksioneve t\u00eb reja agreguese ose tabelare.\n
\u00cbsht\u00eb shtuar mb\u00ebshtetja p\u00ebr sistemet me bibliotek\u00ebn standarde C Musl. <\/ul>\n