1
I Use This!
Very High Activity

Commits : Listings

Analyzed about 16 hours ago. based on code collected about 17 hours ago.
Aug 19, 2024 — Aug 19, 2025
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
tests: now validate AesNiHash128() against reference vectors on all platforms - we have now some reproducible regresssion tests running on both i386 and x86_64 More... 29 days ago
core: fixed / full rewrite of AesNiHashXmm0 to be truly 128-bit - the GoLang reference code was truncated the seed and the result to 64-bit - the 128-bit seed is fully used now (GoLang used a 64-bit seed) - previous implementation did wrongly return highest 64-bit of hash filled with 0 for buffers > 128 bytes, which was just fine on GoLang purpose returning 64-bit results, but wrong about our expectations - x86_64 asm now follows the same exact algorithm than i386 - no performance penalty noticeable, even if x86_64 loop is smaller (still faster than other algorithms including crc32c) - less overhead on Win64 for buffers up to 64 bytes (when XMM6-XMM15 are not used) - shorter, cleaner and more auditable code, also with more useful comments about the algorithm More... 29 days ago
core: fixed long standing issue on i386 AesNiHash() core process - smallest buffers of size 1..16 may be not properly hashed, with potential collisions More... 29 days ago
core: refactored TInterfaceMethod processing functions More... 29 days ago
delphi: fixed weird compiler restriction - the Delphi compiler is not able to properly recognize the expected TDocVariantData.AddValue overload - added new explicit TDocVariantData.AddValueNameLen and TDocVariantData.AddValueJson methods More... 29 days ago
core: simplified XorEntropy() - a single function is enough, since we can't rely on Intel RDTSC availability More... 29 days ago
core: always check for actual RDTSC availability - code should ensure that cfTSC is included in CpuFeatures flags since this opcode may trigger a GPF if CR4.TSD bit is set on hardened systems - see https://tizee.github.io/x86_ref_book_web/instruction/rdtsc.html and https://blog.cr0.org/2009/05/time-stamp-counter-disabling-oddities.html More... 30 days ago
mvc: fixed missing TInterfaceMethodExecuteCached initialization More... 30 days ago
mvc: fixed cookie process independent from TRestServer - we define our own threadvar with the current renderer instance More... 30 days ago
mvc: rewritten core views rendering More... 30 days ago
mvc: rewritten static file process More... 30 days ago
rest: rewritten TRestServerUriContext.ReturnFileFromFolder - and made TRestUriContext.ReturnFileFromFolder abstract More... 30 days ago
orm: refined TRestOrm.AcquireJsonWriter - using a non-resizable 128KB internal buffer makes sense in this context More... 30 days ago
orm: changed RestOrm.GetJsonValue() signature More... 30 days ago
core: refined StartupEntropy content More... 30 days ago
core: some StatusCodeToText() comments precisions More... 30 days ago
core: rewritten _ansistr_concat_multi_convert FPC RTL replacement - to avoid transient memory allocation within 8KB of output More... 30 days ago
all: comment TTextWriterStackBuffer as 8KB work buffer on stack More... 30 days ago
core: new TTextWriter.GetTextAsBuffer method - when allocating a separated RawUtf8 is not worth it - with proper regression tests More... 30 days ago
mvc: fixed TInterfaceMethodExecuteCached usage - those cached instances where not shared between requests, but all created and released for each request! More... 30 days ago
core: enhanced TInterfaceMethodExecuteCached - to publish its TTextWriter as property and not as a separated variable - to include some optional TTextWriter.CustomOptions - ensure the shared instance has a 32KB non-resizable work memory buffer - ensure the non-shared / transient instances (on lock contention) have a fast 2KB sizable buffer More... 30 days ago
tests: validate CookieFromHeaders() More... 30 days ago
core: removed MilliSecsPerSecShl constant - usage of this trick may not be consistent with the exact result of GetTickSec, and may lead to unexpected behavior More... 30 days ago
all: use GetTickSec whenever possible - this is a huge commit for sure, impacting all part of the framework - but it should be safe to use second resolution if possible - and it would avoid any confusion with the RTL / TThread GetTickCount64 overload function/method, which may be incorrectly implemented e.g. on XP More... 30 days ago
core: new GetTickSec function More... 30 days ago
core: added optional crc parameter to DefaultHash/crc32cHash functions More... 30 days ago
tests: refined JWT tests message (to investigate a LUTI error) More... 30 days ago
tests: enhanced TFormatUtf8 coverage More... about 1 month ago
core: deep rewrite of TFormatUtf8 process - simplified and optimized the code especially for AppendLine() and Append()/Make()/MakePath() - FormatUtf8() is not affected More... about 1 month ago
core: introducing TempUtf8Done() - much faster than previous code to release a TTempUtf8 local instance - we know the TempRawUtf8 instance would have a RefCnt=1 so we can just Free the memory if needed More... about 1 month ago