Hungarian translation

Forum Forums antiX-development Translations Hungarian translation

Tagged: 

  • This topic has 40 replies, 7 voices, and was last updated Jul 24-1:56 pm by Kempelen.
Viewing 15 posts - 16 through 30 (of 41 total)
  • Author
    Posts
  • #97284
    Member
    Kempelen

      @Robin: thank you for the further instructions.

      Sorry, I skipped the header, I skipped to “Workflow and Instructions” and missed your note about tx versions. 🙂 Maybe you could move it to step 1 where it says Install and set up transifex client, that’s where I immediately headed to Synaptic and tried to search for it and was happy to see this exists as a deb.

      The token: yes, thank you, I did that yesterday, that’s why my code output above starts with an OK to “Verifying Token…”

      One thing I forgot yesterday, line 75 has a typo that also results in errors if someone copies your example command. There is “anitx” instead of “antix” (but I noticed that soon)

      /bin/bash po-ĂŒbersetzung.sh antix-linux-community-contributions anitx-contribs en@pirate unplugdrive en

      I’ve upgraded my “tx”. Unfortunately I still get an error message:

      po-ĂŒbersetzung.sh antix-linux-community-contributions antix-contribs en@pirate unplugdrive hu
      Keine .pot Datei gefunden. Hole aktuelle .pot-Datei vom transifex Server
      # Getting info about resources
      antix-contribs.unplugdrive - 404, not_found: No <code>Project</code> resource found with the identifier <code>o:antix-linux-community-contributions:p:antix-contribs</code>
      [##############################] (1 / 1)
      Aborted
      /home/lion/bin/po-ĂŒbersetzung.sh: line: 174: ./po/unplugdrive.pot-8678: No such file or directory
      [....]

      Then it continues with some more file not found errors. But if I want to download the PO files, and not do machine translation, maybe I could continue with “tx” only. (Before I opened your script, I didn’t even know this “thing” exists! 🙂 )

      Thanks for the other answers too, that’s great!

      #97285
      Member
      Kempelen

        I did pull all HU files, so now I can make a backup, which should allow me to compare further contributions by others. This is excellent, thanks again! 🙂

        
        tx init
        tx add remote --file-filter 'translations/<project_slug>.<resource_slug>/<lang>.<ext>' http://www.transifex.com/antix-linux-community-contributions/antix-contribs/dashboard/
        tx pull -l hu
        #97354
        Member
        Robin

          Many thanks, Kempelen, for spotting my typo in line 75, I’ll fix it in next update soon, and also follow your proposal to mention the need not to use tx version from debian repo to the step by step instructions.

          I’ve upgraded my “tx”. Unfortunately I still get an error message:

          This most probably happens because you’ve had installed the outdated tx client before on your system.

          After replacing the outdated python tx client by the new go tx client you need to update your ~/.transifexrc file. E.g. it needs to contain an additional line

          rest_hostname = http://rest.api.transifex.com

          to avoid the 404 error messages. These cause the followup file not found errors; as described in the translation script instructions my script doesn’t contain any checks or error handling by now.

          Please check whether this line is present. If this line is not present, you need to run
          tx migrate
          once. Then the error message about resource not found should not come up any longer.

          See Transifex developers site: Migrating from older versions of the client

          which should allow me to compare further contributions by others.

          Well, now I have a question: How do you spot differences within the translated strings? You’ll notice soon, that two .po files can contain 100% identical translation strings without being identical itself as a file. This is due to the fact you can split up any entry into multiple lines (as described in the instructions inside my translation script)

          Example:
          This entry
          msgstr "Az aktuĂĄlis program elƑnĂ©zetĂ©nek megjelenĂ­tĂ©se a böngĂ©szƑben"

          is 100% identically to that entry:

          msgstr "Az aktuĂĄlis program"
          " elƑnĂ©zetĂ©nek megjelenĂ­tĂ©se "
          "a böngĂ©szƑben"

          and also 100% identically to:

          msgstr ""
          "Az "
          "aktuĂĄlis program"
          " elƑnĂ©zetĂ©nek"
          " megjelenítése "
          "a böngĂ©szƑben"

          Same goes for all msgid entries.

          These kind of reformatting is done sometimes by transifex silently and without notice, as well as by the local program poedit, which is the reason why you can’t rely on a simple checksum comparison to compare two versions of a .po file translation. You can’t even compare it using tools like diff or meld, which will show you hundreds or thousands of differences, while actually no difference in translation exists.

          The proper comparison of two versions of a translation is an important task, i.e. when we talk about huge resource like e.g. zzzFM. You simply can’t look up a tiny difference manually in it’s po file.

          Single way out of this is to parse the .po files, extracting all the msgstr entries to single line entries and compare them individually to the respective entry from the other version file.

          Any ideas? I believe a proper comparison could be done by a script, similar to my translating script, but reading both files entry by entry, comparing them rather than translating. This would allow to check all the translated resources more frequently for changes, allowing to check what actually has been changed since last update in the repository.

          Windows is like a submarine. Open a window and serious problems will start.

          #97358
          Member
          Kempelen

            Hi @Robin

            I did “tx migrate” yesterday, before I posted my reply. Without that nothing worked. (E.g. you can see, I can download the whole project.) So the rest hostname is there already (I double checked now). So the error 404 lies somewhere else. Please note that the error says:

            No Project resource found with the identifier o:antix-linux-community-contributions:p:antix-contribs

            Is that a correct reference to “project resources”? I cannot spot any more typos in it, but there might be some error with that o:…p:… thing. And I did completely delete the folders that your script created, to make a fresh start (that was yesterday too). I’m sorry if this is still my “user error”. 🙂

            Regarding comparison: since I didn’t try yet, I cannot really tell. I have two ideas in advance:

            I will commit yesterday’s backup to a local git repository. Then with any future update I can do a git diff – or use a visual git diff program. If strings do not shuffle up and down in high numbers, this should be easy to overview. What I will want to spot is total failures in translations, from contributors who clearly have no clue about what the text means. Those should be easy to spot. And then I fix them on Transifex. Download again, if looks better, commit to git. And leave it for another time period.

            The line-break change you mentioned: I could easily skip those in the DIFF VIEW, if they don’t happen in large quantities.

            My plan B is to use Lokalize’s PO file comparison functions. I have not tried this yet, but I heard some other Hungarian translators use this to find new translations to overview others’ changes. It can compare two PO files visually, and I hope that does all the file parsing correctly.

            I just need to skim over all changes, and if updated texts look “reasonable”.

            Most errors in antiX original HU translation were so bad, that they are easy to spot. E.g. “Firefox localization for Hungarian” was something like “Firefox positioning for Hungarian” (location, rather than localization). Or “Midnight Commander” was translated to “ÉjfĂ©l Parancsnok”. I can easily spot if the translator knew what he is doing or not, and that is my aim. Not perfect in-context checking of everything. The current (original antiX 22) translation is so bad, it’s painful to open in Hungarian. (Fortunately it was 5% translated only. 🙂 )

            I maintain some other translations too, in software that allows seeing “list of changes”. On those I usually just read the change list quickly. (Mouse over displays the English text in CrowdIn if I want to verify better.) Usually the errors are “typical”, e.g. translating a term that I pre-defined to something else – I spot those easily. Or using Friendly language instead of Formal. If the translation looks as a perfect Hungarian sentence, I usually skip to next. Or if I suspect, that the sentence had such complexity as it could be misunderstood by the translator, then I double check with “mouse over”. (Some sentences are very tricky – that’s where machine translation bleeds out too, but some Hungarians fail more often than Deepl.com! :-))) (Especially if they don’t know the specifics of the software. E.g. Live USB things in antiX. I did try all the Remaster and other things on Pendrive before even starting translation. I have to pendrive with antiX. And I have the full install, both x86_64 on my PC, and 32bits on old laptop. And I’m testing the installer too (Live->Install) in VirtualBox that is. I do that in Ubuntu because I cannot boot the antiX Live in antiX VirtualBox. 🙂 I took screenshots of all that, so I can look back any of the original English strings without installing or booting. Maybe I should upload those screenshots somewhere, so other translators can see too. But I have OCR-ed their text, and I want to make a program so that text can be searched somehow… (Even if only with CTRL-F in the browser.) (OCR does not work well on boot process, but it works OK in normal GUI programs.))

            Thanks for your help!

            #97361
            Member
            Robin

              So the error 404 lies somewhere else.

              Strange thing. Just checked and ran it on a couple of languages from here:
              (as you can see, it properly checks and skips the already completely translated German language before starting translation of the empty strings of Afrikaans.)

              demo@antix1:/media/sda4/Arbeitsordner/Übersetzung
              $ /bin/bash po-ĂŒbersetzung.sh antix-linux-community-contributions antix-contribs-testing en@pirate recently-used en
              Keine .pot Datei gefunden. Hole aktuelle .pot-Datei vom transifex Server
              # Getting info about resources
              antix-contribs-testing.recently-used - Done
              [##############################] (1 / 1)
              # Pulling files
              antix-contribs-testing.recently-used [source] - Done
              [##############################] (1 / 1)
              Kein Übersetzungsmuster gefunden. Bitte erstellen Sie die Datei
                ./antix-contribs-testing/recently-used/po/BEARBEITET.recently-used.pot
              aus der original .pot Datei und starten das Script erneut.
              

              Now copy ./antix-contribs-testing/recently-used/po/recently-used.pot to ./antix-contribs-testing/recently-used/po/BEARBEITET.recently-used.pot and repeat the command from before.

              
              demo@antix1:/media/sda4/Arbeitsordner/Übersetzung
              $ /bin/bash po-ĂŒbersetzung.sh antix-linux-community-contributions antix-contribs-testing en@pirate recently-used en
              3352655 (process ID) old priority 0, new priority 14
              Lese .pot-Datei: 52 Zeilen.
              --------------------
              *  Übersetze zu de
              --------------------
              Hole Übersetzung vom transifex-Server
              # Getting info about resources
              antix-contribs-testing.recently-used - Done
              [##############################] (1 / 1)
              # Pulling files
              antix-contribs-testing.recently-used [de] - Done
              [##############################] (1 / 1)
              Sprache ist bereits zu 100% ĂŒbersetzt. Wird ĂŒbergangen.
              Erzeuge Dateien im git-Verzeichnis.
              # Getting info about resources
              antix-contribs-testing.recently-used - Done
              [##############################] (1 / 1)
              # Pulling files
              antix-contribs-testing.recently-used [de] - Done
              [##############################] (1 / 1)
              --------------------
              *  Übersetze zu af
              --------------------
              Hole Übersetzung vom transifex-Server
              # Getting info about resources
              antix-contribs-testing.recently-used - Done
              [##############################] (1 / 1)
              # Pulling files
              antix-contribs-testing.recently-used [af] - Local file is newer than remote, skipping
              [##############################] (1 / 1)
              # Getting info about resources
              antix-contribs-testing.recently-used - Done
              [##############################] (1 / 1)
              # Pulling files
              antix-contribs-testing.recently-used [af] - Done
              [##############################] (1 / 1)
              Verarbeite po-Datei, 50 Zeilen
              Recent files Dynamic menu
              Onlangse lĂȘers Dinamiese spyskaart
                Are you sure you want to delete your Recent files history?  
              Is jy seker jy wil jou Onlangse lĂȘers-geskiedenis uitvee?
              Delete
              Vee uit
               Done!
              Klaar!
              Recent Files to Display
              Onlangse lĂȘers om te vertoon
               Enter the number of recent files  \\n you want to be displayed
              Voer die aantal onlangse lĂȘers in wat jy wil vertoon
              # Getting info about resources
              antix-contribs-testing.recently-used - Done
              [##############################] (1 / 1)
              # Pulling files
              antix-contribs-testing.recently-used [af] - Done
              [##############################] (1 / 1)
              Übertrage komplettierte Übersetzung zum transifex Server
              # Getting info about resources
              antix-contribs-testing.recently-used - Done
              [##############################] (1 / 1)
              # Pushing translations
              antix-contribs-testing.recently-used [af] - Done
              [##############################] (1 / 1)
              Erzeuge Dateien im git-Verzeichnis.
              # Getting info about resources
              antix-contribs-testing.recently-used - Done
              [##############################] (1 / 1)
              # Pulling files
              antix-contribs-testing.recently-used [af] - Done
              [##############################] (1 / 1)
              *  Übersetzung zu af abgeschlossen. Fortsetzung in 3 Sekunden. Taste »b« zum Beenden.
              --------------------
              *  Übersetze zu am
              --------------------
              Hole Übersetzung vom transifex-Server
              # Getting info about resources
              ...

              I’ve added you to antiX-contribs-testing team on transifex, so you can use this additional testing space for what it is meant: testing.
              In order to check out why it didn’t work for you, please try in roxterm (cd to the working folder where the script and the sprachen.lst file is located first):

              $ /bin/bash po-ĂŒbersetzung.sh antix-linux-community-contributions antix-contribs-testing en@pirate recently-used en

              This should skip all languages translated already and start translating all the untranslated strings of the resource “recently-used” in testing space.
              Btw, it is a good idea to copy and paste the transifex organisation-“slug”, the project-“slug” and the resource-“slug” directly from the URL line displayed in browser, e.g.:

              http://www.transifex.com/antix-linux-community-contributions/antix-contribs-testing/recently-used/

              As you can see, mostly everything needed is present therein.
              In antiX-contribs and antiX-contribs-testing the project language is en@pirate (in order to allow all foreign languages a program might be written in originally as source language), while the resource “recently-used” has “en” as true source language obviously.

              Btw, from what OS are you running the script? Maybe there is some difference in the shell or some other tool differs? Remember, I’m not a programmer and have no clue how to make sure a script can be written to be portable. So it may easily work only on antiX. (originally written on antiX 17, but the recent version was only checked on antiX 21/22.)

              Windows is like a submarine. Open a window and serious problems will start.

              #97397
              Member
              Kempelen

                Hi @Robin,

                Thank you. I don’t have access to that “testing” project, it says

                lion@kmp:~/backup/antix_transifex$ /bin/bash po-ĂŒbersetzung.sh antix-linux-community-contributions antix-contribs-testing en@pirate recently-used en
                Keine .pot Datei gefunden. Hole aktuelle .pot-Datei vom transifex Server
                # Getting info about resources
                antix-contribs-testing.recently-used - 403, permission_denied: You do not have permission to perform this action.

                My Transifex username is “antixhu“.

                I tried to copy-paste another program’s all “slug” parts from an URL, as you recommended, but still nothing:

                http://www.transifex.com/antix-linux-community-contributions/antix-contribs/antixscreenshot2/

                lion@kmp:~/backup/antix_transifex$ bash po-ĂŒbersetzung.sh antix-linux-community-contributions antix-contribs en@pirate antixscreenshot2 hu
                Keine .pot Datei gefunden. Hole aktuelle .pot-Datei vom transifex Server
                # Getting info about resources
                antix-contribs.antixscreenshot2 - 404, not_found: No <code>Project</code> resource found with the identifier <code>o:antix-linux-community-contributions:p:antix-contribs</code>
                [##############################] (1 / 1)
                Aborted
                /home/lion/bin/po-ĂŒbersetzung.sh: sor: 174: ./po/antixscreenshot2.pot-6262: No such file or directory

                Please note that I tried with both “hu” and “en” as last parameter, same result. And I tried re-downloading your program, just to make sure I didn’t break it while reading it.

                I’m on antiX 22. (I didn’t know antiX before this version. 🙂 )

                I think I followed all your instructions. But since I’m a programmer, and I know bash too, I should be able to solve this instead of asking you. I just hoped it’s something simple that you notice immediately! We can leave this for now, and I’ll get back to it later. I would like to dedicate my “antiX time” to translation, I’m so near to 100% on the main repo, only a few MX stuff are missing! (And the hundreds of “Firefox localization for XY”, “LibreOffice localization for XY” in antix-packageinstaller-descriptions – I don’t plan to translate those. Or maybe I’ll use automation on those using your program. 🙂 )

                Thanks again, I’ll let you know if I figure out what was wrong.

                #97398
                Member
                Robin

                  My Transifex username is “antixhu“.

                  OK, this makes things clear. I added an existing user “kempelen (F… V…)” at transifex to the contribs-testing team, thinking this was you.
                  Now I’ve added antixhu also. If the kempelen account on transifex doesn’t belong to you, please let me know, then I’ll remove it from the testing team again immediately.

                  So you can easily check out now with antixhu whether the testing line on contribs-testing I provided above works without producing this strange 404.
                  Strange, because the script creates all the needed local structures and its hidden .tx subfolders containing the config for each resource on its own, and here it works fine for me always (I re-downloaded the script from gitlab for testing). So I can’t think of any reason why it should fail for you. By any chance, have you changed your default IFS in shell?

                  Btw, hu is wrong here, it would be hu only if true source language of translatable strings in a resource would be hu instead of en. But this parameter is only used to tell translate-shell what source language identifier is to be submitted to google, it should not produce any error before actually some string is actually sent in for translation to the server of translation provider. You’ll notice some of our resources need “pt” here, some need “de” and some “fr”. But most of them need simply “en” as true source language. You need to check out the source language in pot file (could probably read out this piece of information from .pot file header automatically, will try to add this feature some day).

                  To make sure no download error happened: The sha-256 checksum of the 0.68 version of the script reads
                  fd94b2f42476401b46d9066d131f3ceda1125221fdc41451e891542e085e9896
                  (sorry, had not put the checksum file on gitlab for this resource by now)

                  Last idea: Maybe adding the line
                  # -*- coding: utf-8 -*-
                  on second line within the script, directly below the shebang, makes a difference?

                  I would like to dedicate my “antiX time” to translation

                  That’s great. When starting translation at antiX-contribs, would you mind to start with aCSTV and antiXscreenshot2 resource? aCSTV will be the very next I plan to package finally, and it would be great if checked Hungarian translation could be included. The next one planned for packaging will be antiXscreenshot2 then. No fixed deadlines, since we do all this in spare time volunteer.

                  Windows is like a submarine. Open a window and serious problems will start.

                  #102805
                  Member
                  Kempelen

                    @Robin Hi again,

                    I would like to opt-out from automated translation for Hungarian language. I need to make too many fixes, and the bad sentences confuse me. And it’s more difficult to keep track of what I already worked on.

                    I’m working on zzzfm now (started the contribs repo with this, because I use this program daily). I need to change about half of the texts that you created with machine translation. 🙁 By quick look it looked better earlier. Some examples in diff format:

                    
                    -msgstr "BĂĄrmi"
                    +msgstr "BĂĄrmikor"
                    -msgstr "A kivĂĄlasztott dĂĄtumtartomĂĄnyon belĂŒl"
                    +msgstr "A kivĂĄlasztott idƑpontok közt"
                    -msgstr "<b>Fåjl név:</b>"
                    +msgstr "<b>FĂĄjl neve:</b>"
                    -msgstr "<b>Keresési helyek</b>"
                    +msgstr "<b>Keresés helye</b>"
                    -msgstr "Szabålyos kifejezéssel"
                    +msgstr "Regulåris kifejezés"
                    -msgstr "<b>FĂĄjltartalom</b>"
                    +msgstr "<b>FĂĄjl tartalma</b>"
                    -msgstr "<b>Fåjlméret</b>"
                    +msgstr "<b>Fåjl mérete</b>"
                    -msgstr "UgrĂĄs mappĂĄra"
                    +msgstr "UgrĂĄs a mappĂĄra"
                    -msgstr "NyĂșjtsd"
                    +msgstr "KinyĂșjtĂĄs"
                    -msgstr "Elfér"
                    +msgstr "Illeszkedés"
                    -msgstr "Központ"
                    +msgstr "Középre"
                    -msgstr "ZoomolĂĄs"
                    +msgstr "NagyĂ­tĂĄs"
                    -msgstr "MĂĄsolat (Ctrl+HĂșzza)"
                    +msgstr "MĂĄsolĂĄs (Ctrl+HĂșzza)"
                    -msgstr "Mozog (Shift+HĂșzza)"
                    +msgstr "ÁthelyezĂ©s (Shift+HĂșzza)"
                    -msgstr "Egyetlen kattintĂĄssal megnyitja a fĂĄjlokat"
                    +msgstr "FĂĄjlok megnyitĂĄsa egy kattintĂĄssal"
                    -msgstr "PĂĄrna:"
                    +msgstr "Helyköz:"
                    -msgstr "Vålassz egy kép fåjlt"
                    +msgstr "Vålasszon egy képfåjlt" 
                    

                    Most of the fixes are either: the text is not suitable to that given place in the program (and usually would be rarely good in any program), OR the text is not using an usual wording for software. Or in some rare cases machine created “friendly” grammar, but I use “formal”.

                    E.g. interesting examples from above: “Copy”: “MĂĄsolat” -> “MĂĄsolĂĄs”. The original meaning is a physical thing that exits. E.g. if you made a copy of a paper, so that’s a copy (“MĂĄsolat”). While my translation is the act of copying (“MĂĄsolĂĄs”). (It’s in the drag&drop configuration of ZZZFM). Then “Move”: “Mozog” -> “MozgatĂĄs” – the machine translation means something that is moving by itself. E.g. a file starts moving on my screen to somewhere else. 🙂 “Mozog” is “Moves” in English.

                    Thank you for the work, of course! It’s about 50% good, but for completing the other 50% I think it’s not worth the time save. Makes my time waste on trying to figure out what’s written there in Hungarian and what it should be. 🙂

                    Best regards,
                    Kempelen

                    #102806
                    Member
                    Kempelen

                      And as usual I forgot something: do you think this ZZZFM translation goes to “upstream” zzzfm too, or just antiX? I would like it to be included in all zzzfm releases.

                      #102823
                      Member
                      marcelocripe

                        Thank you for the work, of course! It’s about 50% good, but for completing the other 50% I think it’s not worth the time save. Makes my time waste on trying to figure out what’s written there in Hungarian and what it should be. ????

                        You failed to understand the real intent of machine translations. If you master the English language, then translate the text from the English language to the Hungarian language and modify the automatic translation text that is not consistent with what you expect.

                        If machine translations are 50% good, then they have achieved their goal of being understandable by a native speaker, even if they are 50% bad.

                        Machine translations serve to attract the attention of native speakers and who knows how to make people become volunteer translators, just like you did.

                        Write down the entry or line number you translated so you can pick up where you left off. This is how I do it for the “pt_BR” language.

                        I don’t read or speak English and what I can do is compare machine translations from different services to try to build a coherent text in my language. I compare it to machine translations from several languages other than English, when the original text is incomprehensible.

                        Always give preference to translating what is in Transifex, you can download the file and translate directly into the “.po” or “.ts” file, just have a single error in the file and you will not be able to send your file back to the Transifex and you will waste a lot of time on it.

                        The zzzFM was born in this community, by the hands and enormous intelligence of Skidoo is no longer with us.

                        – – – – –

                        Thank you for the work, of course! It’s about 50% good, but for completing the other 50% I think it’s not worth the time save. Makes my time waste on trying to figure out what’s written there in Hungarian and what it should be. ????

                        VocĂȘ nĂŁo conseguiu compreender a real intenção das traduçÔes automĂĄticas. Se vocĂȘ domina o idioma InglĂȘs, entĂŁo traduza o texto do idioma InglĂȘs para o idioma HĂșngaro e modifique o texto da tradução automĂĄtica que nĂŁo estiver coerente com aquilo que vocĂȘ espera.

                        Se as traduçÔes automåticas são 50% boas, então elas atingiram o seu objetivo que é ser compreensível por um nativo, mesmo estando 50% ruins.

                        As traduçÔes automĂĄticas servem para chamar a atenção dos nativos e quem sabe fazer com que as pessoas se tornem tradutores voluntĂĄrios, assim como ocorreu com vocĂȘ.

                        Anote a entrada ou o nĂșmero da linha que vocĂȘ traduziu, assim vocĂȘ continua da onde parou. É assim que eu faço para o idioma “pt_BR”.

                        Eu nĂŁo leio ou falo em idioma InglĂȘs e o que posso fazer Ă© comparar as traduçÔes automĂĄticas de serviços diferentes para assim tentar construir um texto coerente em meu idioma. Eu comparo com as traduçÔes automĂĄticas de vĂĄrios idiomas diferentes do InglĂȘs, quando o texto original estĂĄ incompreensĂ­vel.

                        Sempre dĂȘ preferĂȘncia em traduzir o que estĂĄ no Transifex, vocĂȘ pode baixar o arquivo e traduzir direto no arquivo “.po” ou “.ts”, basta ter um Ășnico erro no arquivo que vocĂȘ nĂŁo conseguirĂĄ enviar o seu arquivo de volta para o Transifex e vocĂȘ perderĂĄ muito tempo nisso.

                        O zzzFM nasceu nesta comunidade, pelas mĂŁos e da enorme inteligĂȘncia do Skidoo que nĂŁo estĂĄ mais entre nĂłs.

                        #102824
                        Moderator
                        Brian Masinick

                          Marcelo is an excellent example of a person who has learned from others and has done a great job with the translation team.

                          He listens to the good advice of others and relies on online translation services.
                          Marcelo and I have exchanged information even though I know zero Brazilian Portuguese and he knows zero American English. It’s amazing how well we’re able to communicate. 99% of the time it works.
                          1% of the time we have to overcome “slang”, non standard words and rephrase a sentence or phrase. Ask Marcelo. It’s been working most of the time.

                          --
                          Brian Masinick

                          #102902
                          Member
                          Robin

                            Hello @Kempelen !

                            I’m working on zzzfm now (started the contribs repo with this, because I use this program daily). I need to change about half of the texts that you created with machine translation. ???? By quick look it looked better earlier.

                            On the contrary, this zzzFM reflects a very early translation attempt. zzzFM is one of the first resources translated automatically, so the result is probably pretty poor still which is for three reasons:
                            – It was done in 2021, and as you already have said, at that point in time the Hungarian language google translation didn’t return much useful, but we didn’t know it was not at the same level as the translation from e.g. English to German, which had worked acceptable the same time.
                            – zzzFM is one of the most demanding resources, since many code fragments of unpredictable structure are woven into the strings. Automatic translation failed on them (and probably would nowadays still). So in the translation script code at the time we have cut the sentences in chunks, put these on separate lines and marked the lines containing the code as untranslatable. This method has a severe drawback: google translator sees only small splinters, comprising of meaningless sentence fragments. Unfortunately there is no way to simply cut the code fragments from the sentences in this resource, since this would keep the program from working in translation.
                            – Some of the English original strings in this resource are hard to understand or unclear, at least to non native speakers.

                            So best advice I can give you is: keep the 50% of strings which are usable, and simply dump the rest of the meaningless translations. If a translated sentence found in any resource doesn’t make sense, don’t invest time trying understand it, but just translate it completely anew from the original string.

                            I have to admit that specifically in the zzzFM resource some original strings in English language don’t make any sense also to me, if my recollection is correct. Hence I consider these as untranslatable. Probably only native English speakers can understand them.

                            opt-out from automated translation for Hungarian language

                            This was done the very moment you entered the translation team. No automatic hu translation was done since. But let me append, meanwhile the automatic translations are significantly improved, and if you want to let the base work be done by the translation services, just run the script on a specific resource yourself. (in the sprachen.lst file comment out all languages besides hu then). Let me know if you still need assistance in making it run. Otherwise rely completely on manual translation.

                            E.g. interesting examples from above: “Copy”: “Másolat” -> “Másolás”. The original meaning is a physical thing that exits. E.g. if you made a copy of a paper, so that’s a copy (“Másolat”). While my translation is the act of copying (“Másolás”). (It’s in the drag&drop configuration of ZZZFM). Then “Move”: “Mozog” -> “Mozgatás” – the machine translation means something that is moving by itself. E.g. a file starts moving on my screen to somewhere else. ???? “Mozog” is “Moves” in English.

                            As said, probably with zzzFM you’ve scented out the resource of worst automatic translation ever, for the reasons described above. When a sentence is to be cut in chunks (because of containing non removable code fragments), the automatic translation must fail, it can’t decide then whether the word “copy” is meant as a noun or as a verb in the respective chunk. This is since the English has only a single word for both, while other languages might differentiate this in separated grammatical forms.

                            Best regards
                            Robin

                            Windows is like a submarine. Open a window and serious problems will start.

                            #102903
                            Moderator
                            Brian Masinick

                              Regarding zzzFM (and slimski), both were written by an incredibly intelligent person, who was also very witty, and he probably had a tendency to use American “slang” occasionally. Either “big words or slang” are understandably difficult to understand and translate.

                              I’d be happy to explain or help alter any words that are not easily understood or translated; just show me where they are and I’ll help.

                              --
                              Brian Masinick

                              #102964
                              Member
                              marcelocripe

                                – Some of the English original strings in this resource are hard to understand or unclear, at least to non native speakers.

                                Why am I not surprised to read the automatic translation of this sentence?
                                Kempelen, texts in the English language that do not have a beginning, middle and end, it seems to be natural for native speakers of the language, but this is something totally incomprehensible in my language. There are rare machine translation texts that I don’t need to make some kind of adjustment or add a verb that doesn’t exist in the original sentence in the English language. The lack of the verb in the sentences makes it very difficult to understand in automatic translation.

                                I have to admit that specifically in the zzzFM resource some original strings in English language don’t make any sense also to me, if my recollection is correct. Hence I consider these as untranslatable. Probably only native English speakers can understand them.

                                Several phrases were imprecise in the “pt_BR” version, due to the various ambiguous texts existing in SpaceFM, and zzzFM was a fork made by Skidoo from SpaceFM. Some phrases were explained by Skidoo in the topic I referred you to.

                                As said, probably with zzzFM you’ve scented out the resource of worst automatic translation ever, for the reasons described above. When a sentence is to be cut in chunks (because of containing non removable code fragments), the automatic translation must fail, it can’t decide then whether the word “copy” is meant as a noun or as a verb in the respective chunk. This is since the English has only a single word for both, while other languages might differentiate this in separated grammatical forms.

                                This same situation happens in the Brazilian Portuguese language and we have to adjust it manually.

                                I’d be happy to explain or help alter any words that are not easily understood or translated; just show me where they are and I’ll help.

                                Here’s another great example of what we have in this community where one person is willing to help another.

                                He listens to the good advice of others and relies on online translation services.
                                Marcelo and I have exchanged information even though I know zero Brazilian Portuguese and he knows zero American English. It’s amazing how well we’re able to communicate. 99% of the time it works.
                                1% of the time we have to overcome “slang”, non standard words and rephrase a sentence or phrase. Ask Marcelo. It’s been working most of the time.

                                I am very grateful for you agreeing to talk to me, even knowing that the texts are automatic translations. And that many times, automatic translations are funny or are so irregular that they need to be read several times to be understood.

                                Thankfully, machine translation services exist, if it weren’t for them, I wouldn’t be able to participate in the antiX community and would be condemned to the prison called Windows. If these machine translation services continue to evolve as much as other technologies, in the near future we will have machine translation services at the same level as the C-3PO robot in the Star Wars films made by George Lucas.

                                Thank you very much sir. Brian and Robin for always helping me understand the meanings of terms and phrases in the English language.

                                – – – – –

                                – Some of the English original strings in this resource are hard to understand or unclear, at least to non native speakers.

                                Por que serå que eu não estou surpreso em ler a tradução automåtica desta frase?
                                Kempelen, os textos em idioma InglĂȘs que nĂŁo possuem um começo, meio e fim, ao que parece sĂŁo naturais para os nativos do idioma, mas isso Ă© algo totalmente incompreensĂ­vel em meu idioma. SĂŁo raros os textos da tradução automĂĄtica que eu nĂŁo preciso fazer algum tipo de ajuste ou de acrescentar o verbo que nĂŁo existe na frase original em idioma InglĂȘs. A falta do verbo nas frases dificultam muito a compreensĂŁo na tradução automĂĄtica.

                                I have to admit that specifically in the zzzFM resource some original strings in English language don’t make any sense also to me, if my recollection is correct. Hence I consider these as untranslatable. Probably only native English speakers can understand them.

                                VĂĄrias frases ficaram imprecisas na versĂŁo “pt_BR”, devido aos vĂĄrios textos ambĂ­guos existentes no SpaceFM, sendo que o zzzFM foi uma bifurcação feita pelo Skidoo do SpaceFM. Algumas frases foram explicadas pelo Skidoo no tĂłpico que eu te indiquei.

                                As said, probably with zzzFM you’ve scented out the resource of worst automatic translation ever, for the reasons described above. When a sentence is to be cut in chunks (because of containing non removable code fragments), the automatic translation must fail, it can’t decide then whether the word “copy” is meant as a noun or as a verb in the respective chunk. This is since the English has only a single word for both, while other languages might differentiate this in separated grammatical forms.

                                Esta mesma situação acontece no idioma PortuguĂȘs do Brasil e temos que ajustar manualmente.

                                I’d be happy to explain or help alter any words that are not easily understood or translated; just show me where they are and I’ll help.

                                Aqui estĂĄ mais um excelente exemplo do que temos nesta comunidade, onde uma pessoa estĂĄ disposta a ajudar a outra.

                                He listens to the good advice of others and relies on online translation services.
                                Marcelo and I have exchanged information even though I know zero Brazilian Portuguese and he knows zero American English. It’s amazing how well we’re able to communicate. 99% of the time it works.
                                1% of the time we have to overcome “slang”, non standard words and rephrase a sentence or phrase. Ask Marcelo. It’s been working most of the time.

                                Eu sou muito grato por vocĂȘs aceitarem conversar comigo, mesmo sabendo que os textos sĂŁo traduçÔes automĂĄticas. E que muitas vezes, as traduçÔes automĂĄticas sĂŁo engraçadas ou sĂŁo tĂŁo irregulares que precisam ser lidas vĂĄrias vezes para poder ser compreendida.

                                Ainda bem que os serviços de traduçÔes automåticas existem, se não fossem por eles, eu não teria como participar da comunidade do antiX e estaria condenado a prisão chamada Windows. Se estes serviços de traduçÔes automåticas continuarem evoluindo tanto quanto as outras tecnologias, em um futuro próximo teremos os serviços de traduçÔes automåticas no mesmo nível do robÎ C-3PO dos filmes do Star Wars feitos pelo George Lucas.

                                Muito obrigado sr. Brian e Robin por sempre me ajudarem a compreender os significados dos termos e de algumas frases idioma InglĂȘs.

                                #102990
                                Member
                                Kempelen

                                  Hi all,

                                  Thank you for the very friendly responses. 🙂

                                  @marcelocripe: Those “wrong” 50% are also translated to something meaningful usually. The problem with them is that I need to carefully read them, verify that they mean the same thing as the original, and THEN think about whether it could be better said. So this gives many extra steps and slows me down.

                                  To give you an example “device driver”, often referred as “driver” is often translated as the “drive” word in “hard disk drive” -> “MeghajtĂł” in Hungarian. Sometimes this is also used for “device driver” in Hungarian, so sometimes I keep it as it. But there is a better word! “IllesztƑprogram” and most translations use this. So I prefer to use this too, so that it cannot be confused with a “hard disk” – “drive”. (Which is “MeghajtĂł”.)

                                  So often I read the text, I consider it OK and continue on. While it’s not the best possible solution and may become confusing to users.

                                  I have another interesting example. The help page of “zzzfm -g” command was translated by the machine using two different formats:

                                  In one format, it says what the command line switch does.

                                  In another format, it says what the user will achieve by using the command line switch.

                                  Both are correct, and both are good. So by just reading it, I would leave any of them as they were machine translated. BUT! Using the two formats randomly in one help page creates an ugly result. So often I needed to go back and fix what I already fixed, because I realized it is THE OTHER WORDING. 😉

                                  (None of this happens if there is no machine translation.)

                                  These are just some interesting examples. I still do agree with what you said: machine translation is an excellent tool.

                                  @Brian Thank you. If I don’t understand something in the text, I find it in the program and try it. That should help. 🙂

                                  @Robin That makes a lot of sense!

                                  I remember when I looked at the result, it was not so good. Then later I looked at it again, and it was very-very good. But that was a different program. So yesterday I was sad that the result “looked OK but it is not”. But if zzzFM is older translation, that makes everything clear.

                                  OK, then I’ll just continue as usual. 😉

                                  Thanks for all your help guys.

                                  By the way, I totally love zzzFM. I am very sad to hear that it’s not developed anymore.

                                  • This reply was modified 1 year, 1 month ago by Kempelen.
                                Viewing 15 posts - 16 through 30 (of 41 total)
                                • You must be logged in to reply to this topic.