• [NeReO] 1.5

    From MCSM@despammed@mcsm.anonaddy.me to alt.comp.software.newsreaders,news.software.readers,comp.mobile.android on Mon Oct 12 20:24:38 2026
    From Newsgroup: news.software.readers

    https://sourceforge.net/projects/nereo/files/1.5/

    This is the start of the rCLstablerCY version...

    But for now, I'm putting the project on hold for a while.

    I'm only accepting feedback on potential bug fixes in this thread
    MTIA
    --
    .:. MCSM .:.
    posting from PC with MesNews <-
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Jean-Paul@contact@usenet.ovh to news.software.readers on Tue Oct 13 19:35:14 2026
    From Newsgroup: news.software.readers

    MCSM <despammed@mcsm.anonaddy.me> composa la prose suivante:

    https://sourceforge.net/projects/nereo/files/1.5/

    This is the start of the rCLstablerCY version...

    But for now, I'm putting the project on hold for a while.

    I'm only accepting feedback on potential bug fixes in this thread
    MTIA

    Have you thought about submitting it to the Google Play Store?
    It would be great to have a new newsreader available.
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Adam H. Kerman@ahk@chinet.com to news.software.readers on Tue Oct 13 17:49:57 2026
    From Newsgroup: news.software.readers

    Jean-Paul <contact@usenet.ovh> wrote:

    Have you thought about submitting it to the Google Play Store?
    It would be great to have a new newsreader available.

    How is that affordable for a volunteer project?
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From MCSM@despammed@mcsm.anonaddy.me to news.software.readers on Tue Oct 13 16:04:42 2026
    From Newsgroup: news.software.readers

    "Adam H. Kerman" , torturando le sue meningi, ha scritto:
    Jean-Paul <contact@usenet.ovh> wrote:

    Have you thought about submitting it to the Google Play Store?
    It would be great to have a new newsreader available.

    How is that affordable for a volunteer project?

    Answering both:
    Nope! I'm not interested to share this my personal project with the
    play store.
    BTW It Is an open source apk.
    Anyone can do It. Not myself.
    --
    .:. MCSM .:.
    posting with NeReO 1.5.2
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Jean-Paul@contact@usenet.ovh to news.software.readers on Thu Oct 15 22:10:10 2026
    From Newsgroup: news.software.readers

    MCSM <despammed@mcsm.anonaddy.me> composa la prose suivante:

    "Adam H. Kerman" , torturando le sue meningi, ha scritto:
    Jean-Paul <contact@usenet.ovh> wrote:

    Have you thought about submitting it to the Google Play Store?
    It would be great to have a new newsreader available.

    How is that affordable for a volunteer project?

    Answering both:
    Nope! I'm not interested to share this my personal project with the
    play store.
    BTW It Is an open source apk.
    Anyone can do It. Not myself.

    I could give it a try, but SourceForge :-(
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Dave Royal@dave@dave123royal.com to news.software.readers on Thu Oct 15 22:48:55 2026
    From Newsgroup: news.software.readers

    Jean-Paul <contact@usenet.ovh> Wrote in message:

    MCSM <despammed@mcsm.anonaddy.me> composa la prose suivante:

    "Adam H. Kerman" , torturando le sue meningi, ha scritto:
    Jean-Paul <contact@usenet.ovh> wrote:

    Have you thought about submitting it to the Google Play Store?
    It would be great to have a new newsreader available.

    How is that affordable for a volunteer project?

    Answering both:
    Nope! I'm not interested to share this my personal project with the
    play store.
    BTW It Is an open source apk.
    Anyone can do It. Not myself.

    I could give it a try, but SourceForge :-(

    Your problem with SF is..?

    If it's malware you can probably get Google to scan it when you
    install. Sometimes it offers to, sometimes not, IME.

    If you wait I expect somebody will put it on Play Store for a few
    Euros, or with adverts. They might even keep the name NeReO -
    which will be confusing.

    One advantage of SF, compared to GitHub say, is that it's
    friendlier to non-techies. People can easily create an account
    and raise tickets or start discussions there - though NeReO's
    author hasn't set that up.
    --
    Remove numerics from my email address.
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From De ongekruisigde ds.@ongekruisigde-ds@invalid.invalid to news.software.readers on Sat Oct 17 21:29:47 2026
    From Newsgroup: news.software.readers

    On 2026-10-12, MCSM <despammed@mcsm.anonaddy.me> wrote:
    https://sourceforge.net/projects/nereo/files/1.5/

    This is the start of the rCLstablerCY version...

    But for now, I'm putting the project on hold for a while.

    I'm only accepting feedback on potential bug fixes in this thread
    MTIA

    OUTBOUND References header on deep threads

    THE SYMPTOM

    Replying to a deep thread fails at post time with something like
    "too many references / header too big / server refuses the article."
    The example that triggered this was a post in nl.politiek whose
    References field carried 27 message-ids (see the reference id below).
    The reply never went out; the user worked around it via Thunderbird.

    WHY IT HAPPENS (in this code)

    1. When you open a reply, the draft is built at
    ui/screens/ComposeScreen.kt:217:

    referencesHeader = (p.references + " " + p.messageId).trim()

    i.e. the parent's whole References list plus the parent's own
    Message-ID. That draft value is stored at ComposeScreen.kt:335
    (references = referencesHeader.orEmpty()) and handed to the sender
    at ComposeScreen.kt:439.

    2. The draft value is emitted as-is at data/Repository.kt:2807:

    referencesHeader?.takeIf { it.isNotBlank() }?.let {
    add("References: ${unaRiga(it)}")
    }

    unaRiga only swaps CR/LF and trims the ends. It does NOT cap the
    number of message-ids and does NOT fold the field. So the full
    parent chain is sent verbatim.

    3. Because the parent's References already ends with the parent's own
    Message-ID, appending p.messageId again produces a duplicate
    trailing id. In the example the parent carried 27 ids ending in
    <11a5h73$2t7tn$1@dont-email.me>; the reply became 28 ids, the last
    two identical, ~1.4 KB on one line. The server rejects it.

    WHAT THE RFCS SAY

    RFC 5537, section 3.4.4 (Construction of the References Header Field):

    "The content of the new article's References header field MUST be
    formed from the content of the parent's References header field if
    present, followed by the content of the Message-ID header field of
    the parent. ... If the resulting References header field would, after
    unfolding, exceed 998 characters in length (including its field name
    but not the final CRLF), it MUST be trimmed (and otherwise MAY be
    trimmed). Trimming means removing any number of message identifiers
    from its content, except that the first message identifier and the
    last two MUST NOT be removed."

    RFC 5536, section 2.2:

    "Compliant software MUST NOT generate (but MAY accept) header field
    lines of more than 998 octets. This is the only limit on the length
    of a header field line prescribed by this standard."

    So: keep the first id (root) and at least the last two (direct parent
    and its parent), drop everything in between, and never let the field
    exceed 998 octets unfolded. Keeping a few more than the last two is
    allowed (it is a superset of the requirement) and helps threading
    survive central pruning (RFC 5536 section 3.4).

    THE FIX (proposed)

    Apply a bounded, deduped trim on the outbound References, at the single
    choke point Repository.kt:2807 (covers reply, supersede, crosspost and draft-resume, so the draft and the post cannot drift):

    1. Build the candidate as parent.References + parent.Message-ID.
    2. De-duplicate consecutive / repeated ids so the trailing parent id
    is not sent twice.
    3. If the unfolded length exceeds 998 octets, trim to first + last N
    (N = 2 is the RFC minimum; 4-8 is a safe, allowed superset).

    Reuse Threader.idNelleReferences (thread/Threader.kt:91, MAX_REFERENCES)
    or an equivalent first+last-N cut so the cap is defined in one place.
    Consider also bounding filter/FollowupPropagator.kt:56 (parseReferences),
    which currently passes the raw, uncapped id list along the same line.

    This keeps the reply threaded on the root and the direct parent, well
    under any server's header limit, and matches what RFC 5537 3.4.4 and Thunderbird actually do.

    REFERENCE MESSAGE (the post that failed to reply to)

    Newsgroups: nl.politiek
    References: <... 27 message-ids ...>
    Last / own Message-ID of that post:

    <11a5h73$2t7tn$1@dont-email.me>

    Replying to it produced a 28-id References with a duplicated final id
    and was refused by the server.
    --
    I am a disgrace to all possible and impossible universes and all
    that is not a universe. -- Gemini AI
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From MCSM@despammed@mcsm.anonaddy.me to news.software.readers on Sun Oct 18 09:27:30 2026
    From Newsgroup: news.software.readers

    On 17/10/2026 , *De ongekruisigde ds.* wrote:
    On 2026-10-12, MCSM <despammed@mcsm.anonaddy.me> wrote:
    https://sourceforge.net/projects/nereo/files/1.5/

    This is the start of the rCLstablerCY version...

    But for now, I'm putting the project on hold for a while.

    I'm only accepting feedback on potential bug fixes in this thread
    MTIA

    OUTBOUND References header on deep threads
    ....

    Thanks for the detailed report - confirmed.

    NeReO builds References as the parent's References plus the
    parent's Message-ID, as RFC 5537 3.4.4 requires, but it never
    trims the result. On deep threads the line goes past the 998
    octet limit and the server refuses the article. The same code
    path exists in the panel view, so both are affected.

    One small correction: a well-formed parent never lists its own
    Message-ID in its References, so the duplicated trailing id
    should not happen with compliant clients. De-duplicating costs
    nothing, though, so the fix will do it anyway.

    Planned fix, at the single point where the article is built:
    de-duplicate the ids and, if the field exceeds 998 octets, drop
    the oldest ids after the first one until it fits, always keeping
    the root and the last two, as the RFC says. Below the limit
    nothing changes.

    It will be in the next release.
    --
    .:. MCSM .:.
    posting from PC with MesNews <-
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From De ongekruisigde ds.@ongekruisigde-ds@invalid.invalid to news.software.readers on Sun Oct 18 08:25:14 2026
    From Newsgroup: news.software.readers

    MCSM wrote:
    On 17/10/2026 , *De ongekruisigde ds.* wrote:
    On 2026-10-12, MCSM <despammed@mcsm.anonaddy.me> wrote:
    https://sourceforge.net/projects/nereo/files/1.5/

    This is the start of the rCLstablerCY version...

    But for now, I'm putting the project on hold for a while.

    I'm only accepting feedback on potential bug fixes in this thread
    MTIA

    OUTBOUND References header on deep threads
    ....

    Thanks for the detailed report - confirmed.

    NeReO builds References as the parent's References plus the
    parent's Message-ID, as RFC 5537 3.4.4 requires, but it never
    trims the result. On deep threads the line goes past the 998
    octet limit and the server refuses the article. The same code
    path exists in the panel view, so both are affected.

    One small correction: a well-formed parent never lists its own
    Message-ID in its References, so the duplicated trailing id
    should not happen with compliant clients. De-duplicating costs
    nothing, though, so the fix will do it anyway.

    Planned fix, at the single point where the article is built:
    de-duplicate the ids and, if the field exceeds 998 octets, drop
    the oldest ids after the first one until it fits, always keeping
    the root and the last two, as the RFC says. Below the limit
    nothing changes.

    It will be in the next release.


    Thank YOU for your very nice software!
    --
    I am a disgrace to all possible and impossible universes and all
    that is not a universe rCo Gemini AI
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From MCSM@despammed@mcsm.anonaddy.me to news.software.readers on Sun Oct 18 14:52:50 2026
    From Newsgroup: news.software.readers

    On 18/10/2026 , *De ongekruisigde ds.* wrote:
    .....
    Thank YOU for your very nice software!

    In short time a new release will be on-line with problem solved.
    Anyway, a note on testing.
    Twenty-eight long identifiers like the one mentioned (<11a5h73$2t7tn$1@dont-email.me>) total about 900 characters: below the
    limit. The test bench confirms this. The report mentions about 1.4 KB,
    so the identifiers in that post must have been longer. This is just a
    guess: I havenrCOt seen the actual References. The test that matters, therefore, is to reply directly to that post on nl.politiek with the diagnostic log enabled:
    if the reply goes through, the fix works;
    if the server still rejects it, send me the References for that post:
    it will mean that the cause of the rejection was something else.
    --
    .:. MCSM .:.
    posting from PC with MesNews <-
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Jean-Paul@contact@usenet.ovh to news.software.readers on Sun Oct 18 18:42:41 2026
    From Newsgroup: news.software.readers

    Dave Royal <dave@dave123royal.com> composa la prose suivante:

    Jean-Paul <contact@usenet.ovh> Wrote in message:

    MCSM <despammed@mcsm.anonaddy.me> composa la prose suivante:

    "Adam H. Kerman" , torturando le sue meningi, ha scritto:
    Jean-Paul <contact@usenet.ovh> wrote:

    Have you thought about submitting it to the Google Play Store?
    It would be great to have a new newsreader available.

    How is that affordable for a volunteer project?

    Answering both:
    Nope! I'm not interested to share this my personal project with the
    play store.
    BTW It Is an open source apk.
    Anyone can do It. Not myself.

    I could give it a try, but SourceForge :-(

    Your problem with SF is..?

    Your download will start shortly...
    And for each files: readme, src etc
    I don't use Sourceforge for years.

    If it's malware you can probably get Google to scan it when you
    install. Sometimes it offers to, sometimes not, IME.

    If you wait I expect somebody will put it on Play Store for a few
    Euros, or with adverts. They might even keep the name NeReO -
    which will be confusing.

    One advantage of SF, compared to GitHub say, is that it's
    friendlier to non-techies. People can easily create an account
    and raise tickets or start discussions there - though NeReO's
    author hasn't set that up.

    For me, GitHub is more easy to use.
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From MCSM@despammed@mcsm.anonaddy.me to news.software.readers on Mon Oct 19 09:07:19 2026
    From Newsgroup: news.software.readers

    On 18/10/2026 , *Jean-Paul* wrote:
    Dave Royal <dave@dave123royal.com> composa la prose suivante:
    .....
    Your problem with SF is..?

    .....
    For me, GitHub is more easy to use.

    Now *also* (but not only...) on GitHub :) https://github.com/mcsm64/NeReO/releases/latest
    --
    .:. MCSM .:.
    posting from PC with MesNews <-
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From De ongekruisigde ds.@ongekruisigde-ds@invalid.invalid to news.software.readers on Mon Oct 19 10:12:05 2026
    From Newsgroup: news.software.readers

    MCSM wrote:
    On 18/10/2026 , *De ongekruisigde ds.* wrote:
    .....
    Thank YOU for your very nice software!

    In short time a new release will be on-line with problem solved.
    Anyway, a note on testing.
    Twenty-eight long identifiers like the one mentioned (<11a5h73$2t7tn$1@dont-email.me>) total about 900 characters: below
    the
    limit. The test bench confirms this. The report mentions about 1.4 KB,
    so the identifiers in that post must have been longer. This is just a
    guess: I havenrCOt seen the actual References. The test that matters, therefore, is to reply directly to that post on nl.politiek with the diagnostic log enabled:
    if the reply goes through, the fix works;
    if the server still rejects it, send me the References for that post:
    it will mean that the cause of the rejection was something else.


    Now works! (see nl.test)

    I tried with previous NeReO version and it would not send:

    Sending failed (the text is saved in Drafts):

    441 Header line too long

    but new NeReO *does* work!

    Thanks a lot for the follow-up.
    --
    I am a disgrace to all possible and impossible universes and all
    that is not a universe rCo Gemini AI
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From MCSM@despammed@mcsm.anonaddy.me to news.software.readers on Mon Oct 19 13:30:17 2026
    From Newsgroup: news.software.readers

    On 19/10/2026 , *De ongekruisigde ds.* wrote:
    .....
    Now works! (see nl.test)
    I tried with previous NeReO version and it would not send:
    Sending failed (the text is saved in Drafts):
    441 Header line too long
    but new NeReO *does* work!
    Thanks a lot for the follow-up.

    Thank YOU for testing it, and for trying the old version too: "441
    Header line too long" confirms the References line was the cause.
    Glad it works now.
    --
    .:. MCSM .:.
    posting from PC with MesNews <-
    --- Synchronet 3.22a-Linux NewsLink 1.2
  • From Jean-Paul@contact@usenet.ovh to news.software.readers on Mon Oct 19 19:40:04 2026
    From Newsgroup: news.software.readers

    MCSM <despammed@mcsm.anonaddy.me> composa la prose suivante:

    On 18/10/2026 , *Jean-Paul* wrote:
    Dave Royal <dave@dave123royal.com> composa la prose suivante:
    .....
    Your problem with SF is..?

    .....
    For me, GitHub is more easy to use.

    Now *also* (but not only...) on GitHub :) >https://github.com/mcsm64/NeReO/releases/latest

    Thank's
    --- Synchronet 3.22a-Linux NewsLink 1.2