• src/syncterm/GNUmakefile

    From Deuc¿@VERT to Git commit to main/sbbs/m on Friday, October 10, 2025 19:31:00
    https://gitlab.synchro.net/main/sbbs/-/commit/c8388127fa582a0a385b88da
    Modified Files:
    src/syncterm/GNUmakefile updvers.sh
    Log Message:
    Fix up dpkg

    Fix manpath, update version number of output filename.

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Deuc¿@VERT to Git commit to main/sbbs/m on Tuesday, January 20, 2026 06:31:00
    https://gitlab.synchro.net/main/sbbs/-/commit/73daffab73fc8de4acfa5151
    Modified Files:
    src/syncterm/GNUmakefile
    Log Message:
    Fix MANPREFIX usage for dpkg building

    Reported in IRC by phigz.

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Deuc¿@VERT to Git commit to main/sbbs/m on Tuesday, January 20, 2026 06:34:00
    https://gitlab.synchro.net/main/sbbs/-/commit/4b51d850add13f07c4642cfa
    Modified Files:
    src/syncterm/GNUmakefile
    Log Message:
    Add dpkg prefix to dpkg icons

    Reported by phigz in IRC

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Deuc¿@VERT to Git commit to main/sbbs/m on Sunday, April 12, 2026 20:09:00
    https://gitlab.synchro.net/main/sbbs/-/commit/e4478e1d16664ea6e21370fd
    Modified Files:
    src/syncterm/GNUmakefile ripdiff.py ripper.c
    Log Message:
    This is a combination of 81 commits.

    Massive overhaul of RIP support using Claude and the RIPterm test harness Basically, we went through a collection of over 2,000 RIP files collected
    over years from all over the internet. The majority of them are from ACiD packs, but there's also a lot from door games, BBS menu sets, and RIP
    tutorials as well.

    Almost everything has been touched aside from RIP_SET_PIXEL

    Major notes:
    Fonts updated, expecially the default bitmap fonts. SyncTERM now has a new RIPterm font that is used for ANSI output, but this is NOT the same as the "Default 8x8 Bitmap" font that BGI uses. Both are now correct.

    Flood fill now hits the exact same stack limit as RIPterm and uses the
    same seed point ordering and detection, so even RIPs where the fill
    is incomplete are pixel-perfect.

    A HUGE amount of work on the polygon ordering/degenerate handling/filling

    The RIP command parser appears to exactly match RIPterm, so the same
    syntax fails or is parsed and in the same way.

    Full iterative Cohen-Sutherland clipper, just like RIPterm, which special
    magic for thick lines that are partly outside of the viewport.

    The RIP_RESET "fixed" to not reset nearly enough things, just like in
    RIPterm.

    Handle Vertical Tabs in RIP mode.

    MASSIVELY fix a huge amount of button tweakage. RIPterm tends to
    accumulate errors, now SyncTERM does too!

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Deuc¿@VERT to Git commit to main/sbbs/m on Thursday, April 23, 2026 23:45:00
    https://gitlab.synchro.net/main/sbbs/-/commit/7aa9fdfd2524730a5d5cc059
    Modified Files:
    src/syncterm/GNUmakefile
    Log Message:
    SyncTERM: stop rebuilding xp_crypt_botan3 / xp_tls_botan3 every run

    The botan: delegation target in build/botan.gmake has no prerequisites
    and no recipe that creates a botan sentinel file. Every gmake run
    therefore reruns the recipe, and gmake marks botan as "just remade" Ä
    any target that lists botan as a normal prerequisite rebuilds along
    with it. xp_crypt_botan3.o and xp_tls_botan3.o did exactly that.

    Move botan to the order-only side of the prerequisite list on both
    rules. The recursive Botan check still runs first (fast no-op when
    up to date), but its timestamp no longer invalidates the two C++
    objects. A second back-to-back gmake now produces no Compiling lines.

    Mirrors the cl: / cryptlib: pattern, whose consumers depend on the
    real \$(CRYPT_LIB) file path Ä not on the cl pseudo-target Ä and so
    never hit this bug.

    Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Deuc¿@VERT to Git commit to main/sbbs/m on Wednesday, April 29, 2026 15:10:00
    https://gitlab.synchro.net/main/sbbs/-/commit/f36da754351f196969259bf6
    Modified Files:
    src/syncterm/GNUmakefile
    Log Message:
    SyncTERM: bootstrap DeuceSSH cmake configure at recipe time

    The deucessh recipe assumed configure had already run at parse time, but
    the parse-time probe deliberately skips configure when no system crypto
    is on disk Ä leaving the recipe to "cmake --build" a non-existent build
    tree. Run cmake -S/-B inside the recipe when deucessh.pc is missing so
    the unprobed path (fresh checkout, vendored Botan only) actually works.

    Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net
  • From Deuc¿@VERT to Git commit to main/sbbs/m on Wednesday, April 29, 2026 15:28:00
    https://gitlab.synchro.net/main/sbbs/-/commit/b9eb45b6e93811b478487702
    Modified Files:
    src/syncterm/GNUmakefile
    Log Message:
    SyncTERM: pass PKG_CONFIG_PATH explicitly to parse-time pkg-config

    Some gmake versions (observed on Ubuntu) don't propagate
    `export PKG_CONFIG_PATH := ...` from build/botan.gmake into $(shell)
    calls during the same parse phase, so vendored Botan's pkg-config dir
    isn't seen Ä pkg-config falls back to system search and fails to find
    botan-3. The make-side variable is set correctly, so re-emit it on
    the pkg-config command line for the three botan/libcrypto probes that
    matter (DSSH_CRYPTO_READY, TRAILING_LIBS, BOTAN_CXXFLAGS).

    Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

    ---
    þ Synchronet þ Vertrauen þ Home of Synchronet þ [vert/cvs/bbs].synchro.net