I-OpenAI izothola i-Astral futhi ilethe amathuluzi ayo e-Python ku-Codex

Ikhaya » Python » I-OpenAI izothola i-Astral futhi ilethe amathuluzi ayo e-Python ku-Codex

Isinqumo sika I-OpenAI izothenga i-Astral Kuphawula ushintsho oluphawulekayo endleleni izinkampani ze-AI ezicabanga ngayo ngokuthuthukiswa kwesofthiwe. Esikhundleni sokuthuthukisa amamodeli akhiqiza ikhodi, i-OpenAI ithenga amathuluzi ansuku zonke asetshenziswa ngabathuthukisi bePython kakade, ibeka i-Codex ngokujulile ngaphakathi kwemisebenzi lapho ikhodi ibhalwa khona futhi igcinwa khona.

Ngokwezinto ezibonakalayo, lokhu kusho ukuthi u-UV, uRuff kanye no-ty – izinsizakusebenza eziqhutshwa ngokuthula onjiniyela abaningi izikhathi eziningi ngosuku – zizoba yingxenye yesu elibanzi le-OpenAI le-Codex. Lesi sinyathelo sifika ngesikhathi lapho abasizi bokubhala amakhodi be-AI banda khona, futhi ukuncintisana kokwabelana ngemicabango konjiniyela kuya ngokuya kuba mkhulu.

Ukubuka konke kwesivumelwano kanye nomongo wesu

I-OpenAI Group PBC imemezele izinhlelo zokuthuthukisa i-PBC thola i-Astral Software Inc., inkampani entsha kodwa eseyinethonya kakade ohlelweni lwe-Python. Imibandela yezezimali ayidalulwanga, futhi ukuthengiselana kusavunyelwe ngokomthetho kanye nemibandela yokuvala evamile. Kuze kube yilapho abalawuli besayina, i-OpenAI kanye ne-Astral zizoqhubeka nokusebenza njengezinkampani ezihlukene.

I-Astral, eyasungulwa ngo-2022 yi- UCharlie Marsh, bakhe isethi yamathuluzi e-Python asebenza kahle kakhulu abhalwe nge-Rust. Ukugxila kwabo kulula kodwa kunenjongo: ukwenza izinhlelo zikhiqize kakhulu ngokusheshisa imisebenzi ejwayelekile njengokuphathwa kokuxhomekeka, ukubhala phansi kanye nokuhlola uhlobo. Lawo mathuluzi akhule kusukela ku-zero kuya ku- amakhulu ezigidi zokulanda ngenyanga, kanye nabathuthukisi abayizigidi eziningana abathembele kubo emsebenzini wabo wansuku zonke.

Uma isivumelwano sesiphelile, Ithimba le-Astral lizojoyina iqembu le-OpenAI's CodexI-OpenAI ibilokhu icacile ukuthi ukuthengwa kuhloswe ukusheshisa ukuvela kwe-Codex kusuka ku-generator yekhodi ibe uhlelo olungabamba iqhaza emjikelezweni wokuphila wokuthuthukiswa okugcwele, kusukela ekushintsheni kokuhlela kuya ekugcineni amakhodi amakhulu ngokuhamba kwesikhathi.

Izinqubo zomthetho zisalokhu ziwumthetho ongazilungiswa, kodwa Inhloso ye-OpenAI icacile: qinisa ukubandakanyeka kwayo ekusetshenzisweni kwamathuluzi onjiniyela njengoba izimbangi ezifana ne-Anthropic, i-Google kanye ne-GitHub ziphishekela indawo efanayo ngemikhiqizo yazo yokubhala ikhodi ye-AI.

Amathuluzi asekelwe ku-Astral's Rust ePython

Idumela le-Astral lisekelwe kumathuluzi amathathu asetshenziswa kabanzi – u-UV, uRuff kanye no-ty – konke kubhalwe nge-Rust ukuze kusebenze kahle futhi kusebenze kahle. Ukulawula kwezinga eliphansi kanye nemodeli ye-concurrency ye-Rust kuvumela i-Astral ukuthi ifune ukusheshisa okucishe kube ngu-10x kuya ku-100x uma kuqhathaniswa nontanga bendabuko abasekelwe ku-Python, inzuzo eye yathinta kakhulu onjiniyela abaphethe amaphrojekthi amakhulu.

I-flagship, uv, kokubili umphathi wephakheji ye-Python kanye nomphathi wemvelo yephrojekthi. Ivumela amaqembu ukuthi adonse amaphakheji e-Python omthombo ovulekile, aphathe izindawo ezibonakalayo futhi asebenzise kabusha i-cache yendawo yokuxhomekeka ukuze amaphakheji afanayo angadingi ukulandwa ngokuphindaphindiwe. Lokho kuziphatha kwe-caching kuhumushela ekusethweni okusheshayo, amapayipi e-CI angenamsebenzi, kanye nezimanga ezimbalwa lapho kujikelezwa imishini noma izitsha ezintsha.

Ngaphandle kokufakwa okulula, i-UV iyasiza imisebenzi yokulungisa ejwayelekile Njengokufaka ohlwini amaphakheji uhlelo lokusebenza oluthembele kuwo, ukuvumelanisa izinguqulo kuzo zonke izindawo, kanye nokuthuthukela kubahumushi abasha be-Python. Ngokwenza le misebenzi ibe ngokuzenzakalela, kunciphisa ukungqubuzana obekungadla isikhathi sonjiniyela.

I-Ruff, i-linter kanye ne-formatter yesivinini esiphezulu ye-Astral, igxile kwikhwalithi yekhodi kanye nokuvumelana. Ihlola amafayela omthombo we-Python ukuze ithole ukwephulwa kwesitayela, amaphutha amancane kanye nezinkinga zokufometha, bese iphakamisa ukulungiswa okuqondile. Ngenxa yokuthi yakhelwe ku-Rust, ingahlaziya ama-codebase amakhulu ngokushesha, okwenza kube nokwenzeka ukusebenza kuzo zonke izibopho noma njengengxenye ye-tight feedback loops endaweni.

Ukwengeza uRuff kuyinto ty, ithuluzi elikhethekile lokusebenza ngamacebo ohlobo ku-Python. Lapho abathuthukisi bebhala imisebenzi nezakhiwo zedatha ngezinhlobo, ukungalungi noma ukungaguquguquki kungangena ngokuhamba kwesikhathi. I-ty yakhelwe ukubona izimo lapho izichasiselo zohlobo zingahambisani nokusetshenziswa kwangempela, ukubamba amaphutha okungenzeka ngesikhathi sokusebenza kusenesikhathi futhi kwenze ama-refactor aphephe kakhulu.

Uma kuhlanganiswa, i-UV, i-Ruff kanye ne-ty zimboza ingxenye ebanzi yalokho abathuthukisi be-Python abakudingayo ukuze bagcine amaphrojekthi esesimweni esihle: ukuphathwa kokuthembela okusheshayo, ukuphoqelelwa kwesitayela kanye nokuphepha kohloboUkwamukelwa kwabo kuguqule i-Astral kusuka kuphrojekthi ekhethekile yaba yingxenye eseduze nesisekelo yezindlela eziningi zokusebenza ze-Python zanamuhla.

I-Codex: kusukela kumkhiqizi wekhodi kuya kumhlanganyeli womsebenzi

I-Codex ye-OpenAI yaqala impilo njenge- Umsizi wekhodi we-AI okungakhiqiza imisebenzi, ukuhumusha phakathi kwezilimi nokuphakamisa ukuqedwa. Ngokuhamba kwesikhathi, isifiso sande: I-Codex manje ibekwe njengohlelo olungabamba iqhaza kuyo yonke inqubo yokuthuthukiswa kwesofthiwe, kufaka phakathi uhlelo lokusebenza lwe-Codex oluzimele lwe-macOS, hhayi nje ukukhipha izingcezu zekhodi.

Izilinganiso zokusetshenziswa zibonisa ukubambelela okuqinile. I-OpenAI ibika ukuthi i-Codex inokungaphezu kwalokho abasebenzisi abasebenzayo abayizigidi ezimbili masonto onke, ngokukhula kwabasebenzisi okuphindwe kathathu kanye nokwanda komsebenzi okuphindwe kahlanu kusukela ekuqaleni konyaka. Lezo zinombolo zibeka i-Codex phakathi kwamathuluzi okubhala ikhodi e-AI asetshenziswa kabanzi, kodwa futhi zifihla impi eyinkimbinkimbi yokunakwa konjiniyela abangochwepheshe.

Ukuncintisana sekuqine kakhulu. Ikhodi ka-Anthropic kaClaude ithole ukwethembeka kubathuthukisi abaningi, futhi abahleli abanjengoCursor badonsele kokubili udumo kanye nokuxhaswa ngezimali ekuhlolweni okuhehayo. I-GitHub Copilot kanye ne-Gemini Code Assist ye-Google nabo baphokophela kanzima ukuhlanganisa ngokujulile kuma-IDE kanye namapayipi e-CI, ikakhulukazi ngokusebenzisa Ama-ejenti e-AI ku-VS CodeKulowo mongo, i-OpenAI idinga i-Codex ukuze ibe yinto abathuthukisi abayakhayo nsuku zonke, hhayi nje umuntu ohlakaniphile abazamayo ngezikhathi ezithile.

I-OpenAI iveze umgomo wokuguqula i-Codex ibe yi-ejenti engakwazi ukuhlela izinguquko, ukuguqula ama-codebase, ukusebenzisa amathuluzi, ukuqinisekisa imiphumela nokusiza ekulondolozeni izinhlelo ngokuhamba kwesikhathi. Izinsiza ze-Astral sezivele zihleli ngqo kulezo zindlela zokusebenza. Ngokuba ne-toolchain, i-OpenAI ingaxhuma i-Codex emiyalweni efanayo esetshenziswa ngabathuthukisi ukufaka ukuncika, ukuphoqelela imithetho yesitayela nokuqinisekisa izinhlobo.

Ngokombono we-OpenAI, ukuhlanganisa amathuluzi e-Astral kuyindlela yokubheja ukujula phezu kwezici zobuso: kunokuba kwandiswe lokho i-Codex engakukhiqiza ngesikhathi esisodwa, kumayelana nokufaka i-Codex enkambisweni eyinhloko yentuthuko yansuku zonke.

Ukuzibophezela emibuzweni evulekile - kanye nemibuzo evulekile

Enye yezimpendulo zokuqala ezivela emphakathini obanzi yayibikezelwa: Ingabe amathuluzi e-Astral angahlala evulekile? ngaphansi kobunikazi be-OpenAI? Zombili izinkampani zithathe isinyathelo ngokushesha ukubhekana nalolo daba ezitatimendeni zazo zomphakathi.

I-OpenAI isho lokho i-UV, i-Ruff kanye ne-ty zizoqhubeka nokugcinwa njengemiklamo yomthombo ovulekile ngemva kokuphela kokuthengwa, okuhlanganisa lokhu nalokho okuchaza ngokuthi “ifilosofi yonjiniyela kuqala.” Umsunguli we-Astral uCharlie Marsh ubelokhu eqondile ngokufanayo, ebiza umthombo ovulekile ngokuthi “inhliziyo” yomthelela wenkampani futhi ethi uhlala ubalulekile endleleni iqembu elakha ngayo isofthiwe.

Ku-GitHub, i-UV idonsele amakhulu abanikeli kanye nezibuyekezo ezivamile, kuyilapho uRuff no-ty nabo bethuthukise imiphakathi esebenzayo. Lowo mfutho uyingxenye yalokho i-OpenAI ekuthengayo, futhi inkampani ithe i-Astral izoqhubeka nokwakha amathuluzi ayo emphakathini ngisho noma ukuhlanganiswa okujulile ne-Codex kuqhubeka.

Ngesikhathi esifanayo, kukhona imibuzo engaxazululwanga mayelana nokubusa kanye nesiqondiso sesikhathi esideI-OpenAI noma i-Astral azikaze zibeke izinhlelo ezinemininingwane yamamodeli omnikelo, izakhiwo zokwenza izinqumo, noma ukuthi amaphrojekthi azogcina ukuzimela okungakanani uma esengaphansi kwesambulela se-OpenAI.

Kukhona futhi umongo obanzi: I-OpenAI iyeke ukukhipha amamodeli ayo olimi akamuva njengomthombo ovulekile. Izinhlobo ze-Codex zesikhathi esizayo kulindeleke ukuthi zihlale ziyimfihlo, okuphakamisa amathuba okuthi amakhono amasha akhiwe phezu kwamathuluzi e-Astral azohlala ngaphakathi kwezinhlelo zomthombo ovaliwe, noma ngabe izinsiza eziyisisekelo zihlala zinelayisensi evunyelwe.

Okwamanje, amalayisense ku-UV, Ruff kanye ne-ty – MIT, Apache 2.0 kanye nezinketho ezifanayo ezivumelayo – anikeza umphakathi amalungelo aqinile okushintsha, ukushintsha nokusabalalisa kabushaIndlela ukubusa okusebenza ngayo empeleni, nokho, kuzonquma ukuthi amandla omphakathi amanje ayaqhubeka noma athuthela kancane kancane kwenye indawo ngokuhamba kwesikhathi.

Umthelela ohlelweni lwe-Python kanye nemisebenzi yonjiniyela

Ukutholwa kutholakala ohlelweni lwezilimi lapho I-Python isivele ibusa izizinda eziningi: Ucwaningo lwe-AI, isayensi yedatha, ukuzenzakalela, ama-backend ewebhu nokuningi. Amathuluzi e-Astral abe yingxenye ebalulekile yendlela leyo ecosystem esebenza ngayo ngokusebenza, ikakhulukazi kumaqembu akhathalela ukusebenza kanye nokungaguquguquki.

Ngokuhlanganisa la mathuluzi ne-Codex, i-OpenAI ihlose ukwenza qinisa isikhundla sikaPython ekuthuthukisweni okusizwa yi-AI. Umqondo uwukuthi i-Codex ngeke ibhale ikhodi ye-Python kuphela kodwa futhi izophatha ukuncika kwayo nge-UV, iyigcine ihlanzekile nge-Ruff, futhi iqinisekise ukunemba nge-ty, konke lokhu ngenkathi isebenzisana ngqo nemiyalo efanayo abathuthukisi abayenzayo.

Lelo zinga lokuhlanganiswa lingenza Izindlela zokusebenza ezisizwa yi-AI zizwakala zingokwemvelo kakhuluEsikhundleni sokukopisha nokunamathisela iziphakamiso phakathi kwamathuluzi ahlukene, abathuthukisi bangacela i-Codex ukuthi yenze ushintsho, ivuselele amafayela afanele, isebenzise ama-linters kanye nama-type checkers, futhi ibonise imiphumela - konke ngaphakathi kwendawo eyodwa. Ezinhlanganweni ezinkulu, lokhu kungase kuhunyushwe kube amazinga ahambisanayo kanye nemijikelezo yempendulo esheshayo.

Nokho, kunemiphumela yamasu. Ukuba namathuluzi athandwayo kunikeza i-OpenAI isikhundla esikhethekile ku-Python toolchain, into eyenzeka izimbangi ezifana ne-GitHub Copilot, i-Gemini Code Assist ye-Google kanye nabanye abakwazi ukuphinda kalula ngaphandle kokuthengwa okufanayo noma ukubambisana okujulile. Njengoba amathuluzi e-Astral ehlanganiswa kakhulu ne-Codex, kuzovela imibuzo mayelana nokuthi lelo chain lamathuluzi lihlala lingathathi hlangothi kangakanani kubasebenzisi abathanda abanye abasizi be-AI.

Ababukeli bomphakathi baveze kokubili izinzuzo kanye nezingozi ezingaba khona. Ngakolunye uhlangothi, izikhuthazo ezihambisanayo phakathi komhlinzeki omkhulu we-AI kanye nabanakekeli bamathuluzi abalulekile kungasho izinsiza ezingcono, abanakekeli besikhathi esigcwele abengeziwe kanye nezimpendulo ezisheshayo zeziphazamiso noma izinkinga zokuphepha. Ngakolunye uhlangothi, uma izinqumo zesikhathi esizayo mayelana nezici, okuzenzakalelayo noma ukuhlanganiswa kuqondiswa ngokuyinhloko yizinzuzo ze-Codex, abathuthukisi abazimele bangase bazizwe sengathi izidingo zabo zehla ohlwini lwezinto eziza kuqala.

Kulesi sigaba, okuningi kuncike ekutheni i-OpenAI kanye nethimba le-Astral bahlela kanjani ukuphatha kwabo. Amamodeli okuphatha acacile kanye nokucaca okuqhubekayo mayelana nemephu yemigwaqo cishe kuzobaluleka ekugcineni umphakathi obanzi we-Python uhilelekile.

Ithalente, ukuncintisana kanye nokuguquguquka kwemboni

Ngaphandle kwamathuluzi ngokwawo, ukuthengwa nakho kuyi- ukudlala ithalenteIthimba elincane kakhulu le-Astral kodwa elikhethekile kakhulu lizojoyina inhlangano ye-Codex, lilethe ulwazi ekwakheni isofthiwe yengqalasizinda esebenza kahle kakhulu, egxile kubathuthukisi. Lolu hlobo lobuchwepheshe lufunwa kakhulu njengoba izinkampani ze-AI zizama ukwenza imisebenzi enzima ezindaweni zobunjiniyela zangempela.

Lesi sivumelwano siphinde sihambisane nochungechunge lwe- I-M&A ihamba esikhaleni samathuluzi onjiniyelaEzinyangeni ezimbalwa ngaphambi kwesimemezelo se-OpenAI, i-Anthropic yathenga i-Bun, ithuluzi eliphelele le-JavaScript ne-TypeScript elihlanganisa i-running time, umphathi wephakheji, umgijimi wokuhlola kanye ne-bundler. I-Bun yayisivele iyingxenye ebalulekile ye-Claude Code's stack, futhi ukuthengwa kwe-Anthropic kwaqinisekisa ukuthi ukuthembela okubalulekile kuzohlala kuhambisana eduze nohlelo lwayo lomkhiqizo.

Abahlaziyi baphawule ukuthi kokubili ukuthengwa kugqamisa ububanzi umjaho wezikhali uzungeze ukulawula i-toolchainEsikhundleni sokuncintisana ngama-benchmarks amamodeli, izinkampani zithenga amapayipi angaphansi kwabahleli bonjiniyela kanye nemisebenzi ye-CI. Lokho kwenza kube nzima ngabancintisani ukushintshanisa izingcezu zesitaki ngaphandle kokushintsha amathuluzi athembekile kakhulu.

Abanye ababukeli babona ingozi yokuthi i-OpenAI ingagcina isebenzise ubunikazi bayo be-UV noma amathuluzi ahlobene nayo njengesizathu kulo mncintiswano, isibonelo ngezinzuzo zokuhlanganisa ezicashile ne-Codex. Abanye bathi amalayisense avumelekile kanye nokwamukelwa okuqinile komphakathi kwenza kube nzima ukuvala izinto ngaphandle kokucela amafoloko noma ezinye izindlela.

Kanye nalezi zindlela zokuncintisana kunezinto eziningi ezivamile okufanele zicatshangelwe. UMarsh ubonge obala i-Astral's Abatshalizimali be-Series A kanye ne-Series B, okwenza abanye bacabange ukuthi ukuthengwa kungabavumela ukuthi bashintshe amasheya abo ukuze bathole izikhundla ku-OpenAI - inkampani okukhulunywa ngayo ukuthi icabanga ngezinketho zemakethe yomphakathi esikhathini esizayo esingekude kakhulu.

Yiziphi izinguquko konjiniyela njengamanje

Okwamanje, Ukusetshenziswa kwansuku zonke kwe-UV, i-Ruff kanye ne-ty akukashintshiUkuthengwa akukapheli, futhi i-OpenAI ne-Astral zigcizelele ukuthi izinkampani zizosebenza ngokuzimela kuze kube yilapho zonke izinyathelo zomthetho nezinqubo seziqediwe.

Abathuthukisi basengafaka amathuluzi e-Astral emithonjeni evamile, bafake izinkinga ku-GitHub, futhi banikele ngama-patches njengoba bebelokhu bekwenza. Abanakekeli abakhona basesekhona, futhi akukho zinguquko ezimenyezelwe kumamodeli okunikeza amalayisense noma okusabalalisa.

Ukushintsha okubonakala kakhulu cishe kuzovela uma Ukuhlanganiswa kwe-Codex kuqala ukusebenzaI-OpenAI isikisele ukuthi ama-ejenti e-Codex azokwazi ukubiza amathuluzi e-Astral ngqo - isibonelo, ukuqalisa i-UV ukuphatha izindawo ngesikhathi sokuvuselela okuzenzakalelayo, noma ukusebenzisa i-Ruff kanye ne-ty njengengxenye yenqubo yokubuyekeza eqhutshwa yi-AI.

Uma lokho kuhlanganiswa kuklanywe kahle, onjiniyela bangase babone ukujikeleza okuqinile phakathi kweziphakamiso ze-AI nokuqinisekiswaEsikhundleni sokusebenzisa ngesandla ukuhlola noma ukuhlola uhlobo ngemuva kokwamukela ushintsho lwekhodi kusuka ku-Codex, umsizi angasebenzisa amathuluzi, izixwayiso zangaphandle, futhi aphakamise ukulungiswa ngokulandelana okukodwa. Ngokuhamba kwesikhathi, lokho kungasiza ekuxazululeni enye yezinkinga ezinkulu mayelana nekhodi ekhiqizwe yi-AI: ukuthi kungaba nzima ukuyigcina.

Esikhathini eside, uhlelo lwe-Python luzobheka izimpawu zokuthi amaphrojekthi e-Astral ayaqhubeka yini nokushintsha ngezindlela ezikhonza umphakathi obanzi - kufaka phakathi amaqembu angasebenzisi i-Codex, noma athanda abasizi be-AI abancintisanayo. Ukuzibophezela okukhulunywe ngakho emthonjeni ovulekile kanye neziqinisekiso zomthetho zamalayisense amanje kunikeza isiqinisekiso esithile, kodwa ukwethembana cishe kuzoncika ekutheni imijikelezo elandelayo yezinqumo ezinkulu zezici iphathwa kanjani.

Sekukonke, isinyathelo se-OpenAI sokuthola i-Astral sidonsela ukuthuthukiswa okusizwa yi-AI eduze kwamathuluzi asevele esekela ukusebenza kwe-Python yesimanje. Ngokuhlanganisa amakhono okumodela e-Codex kanye ukusebenza okuphezulu, amathuluzi omthombo ovulekile asetshenziswa kabanzi, inkampani ibheja ukuthi ikusasa lokuhlela lizobunjwa kakhulu yimininingwane evamile yokuxazululwa kokuxhomekeka kanye nokuchaza izinto njengoba kunjalo nangokuthuthukiswa kwemodeli ethatha izihloko.

I-athikili ehlobene:

I-OpenAI yethula uhlelo lokusebenza lwe-Codex oluzimele lwe-macOS ngokulawula ama-ejenti amaningi